[email protected] Paper 37 571-272-7822 Entered: June 30, 2021

UNITED STATES PATENT AND TRADEMARK OFFICE

BEFORE THE PATENT TRIAL AND APPEAL BOARD

INTERNATIONAL BUSINESS MACHINES CORPORATION, Petitioner,

v.

TRUSTED KNIGHT CORPORATION, Patent Owner.

IPR2020-00323 Patent 9,503,473 B1

Before THOMAS L. GIANNETTI, BRIAN J. MCNAMARA, and NABEEL U. KHAN, Administrative Patent Judges.

KHAN, Administrative Patent Judge.

JUDGMENT Final Written Decision Determining Some Challenged Claims Unpatentable 35 U.S.C. § 318

IPR2020-00323 Patent 9,503,473 B1 I. INTRODUCTION A. Background and Summary International Business Machines Corporation (“Petitioner”) filed a Petition (Paper 2, “Pet.”) requesting an inter partes review of claims 1–29 (“the challenged claims”) of U.S. Patent No. 9,503,473 B1 (Ex. 1001, “the ’473 Patent”). Trusted Knight Corporation (“Patent Owner”) timely filed a Preliminary Response (Paper 6, “Prelim. Resp.”). With our authorization, Petitioner and Patent Owner filed additional briefing on issues related to discretionary denial under 35 U.S.C. § 314(a). See Papers 8, 10, 13. On July 9, 2020, upon consideration of the Petition, Preliminary Response, the additional briefing, and the evidence cited, we determined that Petitioner established a reasonable likelihood that it would prevail with respect to at least one of the claims challenged in the Petition and instituted review to determine the patentability of the challenged claims on all grounds. Paper 15 (“Dec. Inst.”), 1. After institution, Patent Owner filed a Patent Owner Response (Paper 19, “PO Resp.”), Petitioner filed a Reply (Paper 24, “Pet. Reply”), and Patent Owner filed a Sur-Reply (Paper 23, “PO Sur-reply”). An oral hearing was held on April 8, 2021, and the hearing transcript is included in the record. Paper 32 (“Tr.”). After the oral-hearing, the parties provided supplemental post-hearing briefing addressing claim construction. Papers 33–36. We have jurisdiction under 35 U.S.C. § 6. This Final Written Decision, issued pursuant to 35 U.S.C. § 318(a) and 37 C.F.R. § 42.73 (2019), addresses issues and evidence raised during the inter partes review. For the reasons that follow, Petitioner demonstrates by a preponderance of

2 IPR2020-00323 Patent 9,503,473 B1 the evidence that claims 1–6, 8–17, and 19–29 of the ʼ473 Patent are unpatentable, but does not show that claims 7 and 18 are unpatentable. B. Related Matters The parties identify the following matter as related to this case: Trusted Knight Corporation v. International Business Machines Corporation, Case No. 3:19-cv-1206 (N.D. Cal.) (“underlying district court litigation”). Pet. 86; Paper 3, 1. C. The ’473 Patent The ’473 Patent, titled “Apparatus, System, and Method for Protecting Against Keylogging ,” is directed to “preventing key logger malware that utilizes form grabbing techniques to steal financial and identity information from users’ browsers.” Ex. 1001, 1:22–24. The ’473 Patent describes that the method includes detecting a browser form submission initiation call event associated with data inputs entered by a user. Ex. 1001, 4:16–18. Confidential data is cleared from the data inputs, thereby protecting against the threat of key logging malware capturing the confidential data. Ex. 1001, 4:20–22. The clearing of the confidential data is performed without requiring detection of the key logging malware. Ex. 1001, 4:22–24. Further, the detecting of a browser form submission initiation call event occurs at a most privileged access level. Ex. 1001, 4:25–26. Figure 1 of the ’473 Patent is reproduced below:

3 IPR2020-00323 Patent 9,503,473 B1

Figure 1 depicts an overview of an environment in which an anti-key logger operates. Ex. 1001, 5:14–17. At the keyboard driver level (100), input is provided by a user, and the anti-key logger (105) functions at this level to protect the inputted keyboard data. Ex. 1001, 5:17–19. The virtual keyboard (110) is the next step in the flow of inputted keyboard data, and it is a common location for a malicious key logger (115) to be present to intercept the inputted data. Ex. 1001, 5:19–22. The operating system (120) receives the inputted keyboard data and passes the data to the application (130) being utilized by the user, which is a location where malicious keyloggers (135) also intercept the inputted keyboard data. Ex. 1001, 5:22– 26. Finally, the application passes the inputted keyboard data to the Internet web server (140) per the user request. Ex. 1001, 5:26–28. Figure 2 of the ’473 Patent is reproduced below:

4 IPR2020-00323 Patent 9,503,473 B1

Figure 2 depicts a diagram of actions in defeating the operation of form grabbing key loggers. Ex. 1001, 4:52–54. The anti-key logger software inserts itself in an application programming interface (“API”) stack last, causing the anti-key logger software to be called first (S200). Ex. 1001, 5:42–44. When a BeforeNavigate1 event is identified (S210), the anti-key logger software confirms its placement in the API stack (S215). Ex. 1001, 5:44–46. When a form submission event occurs (S230), the anti-key logger software identifies all forms on the called web page (S220). Ex. 1001, 5:46– 48. If forms are present, the anti-key logger software connects to each form submission event (S222), the anti-key logger software clears all form inputs marked with INPUT or PASSWORD (S224), and then the event handler clears all passwords (S226). Ex. 1001, 5:48–52. The anti-key logger

1 The ʼ473 Patent describes BeforeNavigate as a form submission initiation call event under . Ex. 1001, 10:30–32.

5 IPR2020-00323 Patent 9,503,473 B1 software provides the user inputted data to the OnSubmit event to the designed receiving party, such as a bank (S240). Ex. 1001, 5:52–54. The anti-key logger software also ensures all password form fields are cleared from the API chain (S235) and thus are unavailable to capture by form grabbing key loggers. Ex. 1001, 5:54–57. D. Priority Chain of the ’473 Patent The ’473 Patent issued from U.S. Patent Application No. 15/207,279, filed July 11, 2016 (“the ’279 Application”). The ’473 Patent claims priority to, and incorporates by reference the content of, each of the following applications: U.S. Patent Application No. 14/709,224, filed May 11, 2015 (“the ʼ224 Application,” also “the Parent Application”), now U.S. Patent No. 9,596,250 (“the ’250 Patent”); U.S. Patent Application No. 13/667,256, filed November 2, 2012, now abandoned (“the Abandoned Application”); U.S. Patent Application No. 12/427,833, filed April 22, 2009 (“the ’833 Application,” also “the Grandparent Application”), now U.S. Patent No. 8,316,445 (“the ’445 Patent”); and U.S. Patent Application No. 61/125,178 (“the ’178 Provisional Application”). Prelim. Resp. 22–23; Pet. 11. The priority chain of the ’473 Patent (reproduced from Patent Owner’s Preliminary Response) is depicted below:2

2 The figure incorrectly shows that Application No. 13/667,256 was abandoned July 30, 2005. The correct date of abandonment is July 30, 2015.

6 IPR2020-00323 Patent 9,503,473 B1 The above figure illustrates the relationship between the ’473 Patent and all the intermediate applications claiming priority all the way back to the ’178 Provisional Application. Prelim. Resp. 23. E. Illustrative Claims Of the challenged claims, claims 1, 11, 22, and 26 are independent. Claims 2–10 depend from independent claim 1, claims 12–21 depend from independent claim 11, claims 23–25 depend from independent claim 22, and claims 27–29 depend from independent claim 26. Claim 1, reproduced below with annotations, is illustrative. 1. [1pre] A method for preventing software key logging executable by a microprocessor, comprising: [1a] installing and maintaining an anti-key lodger [sic] at a most privileged access level for browser events in an Application Programming Interface (API) stack; [1b] detecting, by the anti-key logger, a browser form submission initiation call event associated with data inputs entered by a user, wherein the form submission initiation call event is an OnSubmit call event or a BeforeNavigate call event; [1c] submitting the data inputs to a designated entity; and [1d] clearing, by the anti-key logger, confidential data from the data inputs to protect against the threat of key logging malware capturing the confidential data. Ex. 1001, 13:2–16. F. Prior Art and Asserted Grounds Petitioner asserts that claims 1–29 would have been unpatentable on the following grounds:

7 IPR2020-00323 Patent 9,503,473 B1

Ground Claim(s) Challenged 35 U.S.C. § Reference(s)/Basis 1 1–29 102(b) the ’445 Patent3 2 1–6, 8–17, 19–29 103(a) Waterson4, Ross5 3 1–6, 8–17, 19–29 103(a) Waterson, Ross, Geon6

In addition, Petitioner relies on the Declaration of Dr. Patrick McDaniel (Ex. 1013) and the Reply Declaration of Dr. Patrick McDaniel (Ex. 1024) in support of the asserted grounds of unpatentability. See generally Pet.; Pet. Reply. Patent Owner relies on the Declaration of Dr. Adam Sorini (Ex. 2043) and Trevor Reschke (Ex. 2044) in support of Patent Owner’s Response. See generally PO Resp. II. ANALYSIS A. Level of Ordinary Skill in the Art Petitioner argues that a person of ordinary skill in the art at the time of the invention “would have at least a Bachelor’s degree and two years’ work experience in operating systems, device drivers, or anti-malware software, or equivalent work experience.” Pet. 20 (citing Ex. 1013 ¶ 40). Patent Owner does not indicate that it disputes Petitioner’s definition. See PO Resp. 10. We adopt Petitioner’s description, which we determine to be consistent with the level of skill reflected in the asserted prior art references. See Okajima v. Bourdeau, 261 F.3d 1350, 1355 (Fed. Cir. 2001).

3 Liske, US 8,316,445 B2, Nov. 20, 2012 (Ex. 1003, “the ’445 Patent”). The ’445 Patent is the “Grandparent” of the ’473 Patent. See Part I.D., supra. 4 Waterson, US 7,779,062 B2, Aug. 17, 2010 (Ex. 1009, “Waterson”). 5 Blake Ross et al., Stronger Password Authentication Using Browser Extensions, Proc. Of the 14th USENIX Security Symposium (July 31 – Aug. 5, 2005) (Ex. 1010, “Ross”). 6 Geon et al., US 7,774,595 B2, Aug. 10, 2010 (Ex. 1012, “Geon”).

8 IPR2020-00323 Patent 9,503,473 B1 B. Claim Construction In inter partes reviews, we interpret a claim “using the same claim construction standard that would be used to construe the claim in a civil action under 35 U.S.C. 282(b).” See 37 C.F.R. § 42.100(b). Under this standard, we construe the claim “in accordance with the ordinary and customary meaning of such claim as understood by one of ordinary skill in the art and the prosecution history pertaining to the patent.” See 37 C.F.R. § 42.100(b). Only claim terms that are in controversy need to be construed and only to the extent necessary to resolve the controversy. See Nidec Motor Corp. v. Zhongshan Broad Ocean Motor Co., 868 F.3d 1013, 1017 (Fed. Cir. 2017). 1. Claim Construction in the Underlying Litigation The parties have agreed to the construction of three terms in the underlying district court litigation. See Ex. 2034, 2; Ex. 2026, 1 n.1. The first of these terms is “clearing confidential data,” which the parties agreed means “removing meaning from confidential data.” Ex. 2034, 2. The second term is “designated entity,” which the parties agreed means “intended receiving entity over a network.” Ex. 2034, 2. The third term is “an Application Programming Interface (API) stack,” which the parties agreed means “an API stack, accessed by a web browser application, including web browser and operating system APIs.” Ex. 2026, 1 n.1. In addition to these three agreed terms, the parties disputed the construction of a fourth term in the underlying litigation: “most privileged access level.” The district court construed this term to mean “zero-ring level.” Ex. 2026, 13.

9 IPR2020-00323 Patent 9,503,473 B1 Patent Owner argues, and Petitioner agrees, that the Board should apply these aforementioned claim constructions in this proceeding. PO Resp. 9; Pet. Reply 2. We have reviewed the constructions adopted by the district court and the evidence of record in this proceeding and agree with the district court’s constructions. We, therefore, apply the district court’s constructions in this proceeding. Accordingly, we construe “clearing confidential data” to mean “removing meaning from confidential data,” “designated entity” to mean “intended receiving entity over a network,” “an Application Programming Interface (API) stack” to mean “an API stack, accessed by a web browser application, including web browser and operating system APIs” and “most privileged access level” to mean “zero-ring level.” 2. Maintaining an Anti-Key Logger at a Most Privileged Access Level In its Reply Brief, Petitioner argues that Patent Owner applies several incorrect constructions to various claim terms without providing an express construction for these terms. Pet. Reply 2–5. One such term is “maintaining an anti-key lo[g]ger at a most privileged access level.” Pet. Reply 3–4. At the oral hearing it became clear that the parties’ dispute regarding this term stemmed from a disagreement about the scope of the term rather than the disclosure of the prior art. See Tr. 9:8–11:15, 22:14–28:7. Thus, we ordered the parties to provide supplemental post-hearing briefing addressing whether, in light of the Specification of the ʼ473 Patent, the scope of the “maintaining” term (1) requires the step to be performed automatically without intervening user action, such as rebooting the system, and (2) whether the term precludes verifying that the anti-key logger is at a most privileged access level before taking further steps. Paper 31.

10 IPR2020-00323 Patent 9,503,473 B1 Petitioner argues that the ordinary meaning of the word “maintaining” is not limited to processes that automatically maintain, and thus without an expressly contradicting definition in the Specification or a clear disavowal of claim scope, the “maintaining” term does not require the step to be performed automatically. Paper 33, 1. Petitioner argues that had the inventors of the ʼ473 Patent intended the claims to require automatically maintaining, they could have done so. Paper 33, 2. Petitioner argues that there is only one instance in the Specification where “maintaining is ever discussed” and that this instance contains no discussion or suggestion that the anti-key logger maintains its position automatically. Paper 33, 3 (emphasis removed) (citing Ex. 1001, 5:4–6; 8:48–64). Similarly, Petitioner argues that nothing in the Specification would support narrowing the scope of the maintaining term to preclude “verifying.” Paper 33, 4. Further, Petitioner argues that the dependent claims, which recite “wherein the anti-key logger is placement aware and performs at least one check to confirm that the anti-key logger is called first” support the finding that “maintaining” includes “verification.” Paper 33, 4. Patent Owner argues “the ’473 invention does not require user intervention (such as rebooting) because the invention does not depend on the detection of malware.” Paper 34, 2 (citing Ex. 1001, 3:46–52). Patent Owner cites to independent claims 22 and 26 and dependent claims 2 and 12 as support for its argument that the claimed protection exists without detection of key logging malware. Paper 34, 3. Patent Owner further argues that the claims do not require rebooting because the anti-key logger is “placement aware” and that it “‘renegotiate[s] its location’ to ensure that its protection is not ‘circumvent[ed]’ at ‘any time.’” Paper 34, 4 (citing Ex. 1001, 5:36–40) (alterations in original).

11 IPR2020-00323 Patent 9,503,473 B1 As to whether the maintaining limitation precludes verification that the anti-key logger is at a most privileged access level before taking further steps, Patent Owner focuses on the disclosure of Waterson to argue that verification in Waterson involves checking if a driver file has been altered, and if so, requesting the user to reboot. Paper 34, 4–5 (citing Ex. 1001, 5:58–6:3; 6:4–6:17). Patent Owner argues that “‘checking’ is not ‘maintaining.’” Paper 34, 5. We determine that “maintaining an anti-key lo[g]ger at a most privileged access level” does not require the step to be performed automatically without intervening user action. We agree with Petitioner that the Specification does not limit the term in this way. The primary section where the Specification discusses this term is, as Petitioner points out, related to the discussion of Figure 6, which “illustrates the manner in which an embodiment of the invention functions to maintain its position in the API stack by illustration of its relation to kernel level calls.” Ex. 1001, 8:48–50. Here, the Specification simply states that “Ring 0 is the level with the most privileges,” and that “protection is inserted at this level (660) whereby the protection can determine if an unauthorized kernel level call is being made (670) at which time it is bounced from the API chain.” Ex. 1001, 8:50–57. As can be seen, the Specification explains that the anti-key logger is installed at the Ring 0 level (a most privileged level), but does not further explain or restrict how the anti-key logger is maintained at this level other than to bounce unauthorized kernel level calls. In other sections, the Specification explains that “the software confirms its placement in the API stack,” but does not preclude any user actions in response to this confirmation. Ex. 1001, 5:45–46, 6:1–2. Finally, the Specification also describes that in “one embodiment the software . . . is

12 IPR2020-00323 Patent 9,503,473 B1 placement aware and renegotiates its location in the API stack to ensure there are no other hooks that circumvent the protection at any time.” Ex. 1001, 5:36–40. Again, this description does not preclude user action as part of the process. Patent Owner’s argument that the claimed invention does not depend on the detection of malware does not address whether user actions are precluded from the process of maintaining the position of the anti-key logger at a most privileged level. Instead, Patent Owner argues that the ʼ473 Patent does not require user action, such as rebooting, in order to maintain the position of the anti-key logger. Paper 34, 4. Whether the ʼ473 Patent requires user action does not address whether the ʼ473 Patent precludes user action. In other words, that certain embodiments in the ʼ473 Patent do not require user action does not limit the scope of the claims to precluding such user action. “To disavow claim scope, the specification must contain ‘expressions of manifest exclusion or restriction, representing a clear disavowal of claim scope.’” Retractable Techs., Inc. v. Becton, Dickinson & Co., 653 F.3d 1296, 1306 (Fed. Cir. 2011) (quoting Epistar Corp. v. Int’l Trade Comm’n, 566 F.3d 1321, 1335 (Fed. Cir. 2009)). We also determine that “maintaining an anti-key lo[g]ger at a most privileged access level” does not preclude verifying that the anti-key logger is at a most privileged access level before taking further steps. Indeed, the Specification explains that “the software confirms its placement in the API stack.” Ex. 1001, 5:45–46, 6:1–2. By confirming the placement of the software, Waterson verifies its placement. By focusing on Waterson’s disclosure of checking the driver files, Patent Owner’s argument fails to address whether the ʼ473 Patent precludes verifying the placement of the anti-key logger.

13 IPR2020-00323 Patent 9,503,473 B1 In summary, we construe “maintaining an anti-key lo[g]ger at a most privileged access level” to not require the step to be performed automatically without intervening user action, such as rebooting the system, and to not preclude verifying that the anti-key logger is at a most privileged access level before taking further steps. Based on the arguments presented before us, we determine we need not further construe this term. C. Anticipation by the ’445 Patent Petitioner contends claims 1–29 are anticipated by the ’445 Patent. Pet. 23–52. Anticipation under 35 U.S.C. § 102 requires that each limitation in a claim is found in a single prior art reference, arranged as recited in the claim. Net MoneyIn, Inc. v. VeriSign, Inc., 545 F.3d 1359, 1369 (Fed. Cir. 2008). The disclosure of the element by the reference can be an express disclosure or an inherent disclosure. Verdegaal Bros. v. Union Oil Co. of California, 814 F.2d 628, 631 (Fed. Cir. 1987). 1. Overview of the ’445 Patent As previously discussed, the ’473 Patent claims priority to the filing date of the Grandparent Application, which issued as the ’445 Patent. Pet. 11. As contended by Petitioner, the ’473 Patent and the ’445 Patent have virtually identical specifications. Pet. 8, 15, 33. As such we refer back to the overview of the ʼ473 Patent, which applies to the ʼ445 Patent as well. 2. Analysis of whether the ’445 Patent is Prior Art Petitioner contends the ’445 Patent “Grandparent” is prior art to the ’473 Patent because the intervening Parent Application broke the priority chain. Pet. 24; see also Pet. 25–33. According to Petitioner, the Parent Application was not “similarly entitled to the benefit” of any earlier application in the claimed priority chain because it was filed with a completely new specification, all of which constitutes new matter, and the

14 IPR2020-00323 Patent 9,503,473 B1 claims in the Parent Application were all exclusively directed to the new matter. Pet. 23–25 (quoting 35 U.S.C. § 120 (2018)). Petitioner argues that because the Parent Application “itself claimed only new matter, it was not entitled to claim priority to any earlier patent in its lineage.” Id. at 24. According to Petitioner, the ’473 Patent is not entitled to claim priority to the Grandparent Application, because 35 U.S.C. § 120 only allows a priority claim “on an application similarly entitled to the benefit of the filing date of the first application” and because the Parent Application is not similarly entitled to the benefit of the Grandparent Application, it breaks the chain. Id. at 23–25. As further argued by Petitioner, the ’445 Patent issued more than a year before the filing date of the ’473 Patent, and thus, the ʼ445 Patent is anticipatory prior art to the ’473 Patent. Id. We are not persuaded by Petitioner’s arguments. In order to claim the benefit of an earlier filed non-provisional application under 35 U.S.C. § 120, a later-filed application must comply with several requirements, including: (1) the written description of the earlier filed application discloses the invention claimed in the later filed application sufficient to satisfy the requirements of § 112; (2) the applications have at least one common inventor; (3) the later application is filed before the issuance or abandonment of the earlier filed application; and (4) the later application contains a reference to the earlier filed application. In re NTP, Inc., 654 F.3d 1268, 1277 (Fed. Cir. 2011). The evidence shows that the ʼ473 Patent satisfies all these requirements. The disclosure of the Grandparent Application is nearly identical to the disclosure of the ʼ473 Patent. All other descendant applications in the chain, including the ʼ473 Patent and the Parent Application, incorporate the Grandparent Application by reference. See Exs. 1001, 1003, 1005. The evidence shows that all the applications in the chain

15 IPR2020-00323 Patent 9,503,473 B1 have common inventorship. Ex. 2002, 92–96. The evidence also shows that all of the later-filed applications in the family of the ’473 Patent were filed before the “patenting or abandonment of or termination of” the prior-filed non-provisional application, and thus were co-pending for purposes of 35 U.S.C. § 120. Ex. 2002, 114–122. Finally, the ʼ473 Patent references the Grandparent Application, all the intermediate applications in the ’473 Patent family, and the relationships thereto. Ex. 1001, 1:5–13. Petitioner does not dispute that the invention of the ʼ473 Patent has been disclosed by all applications in the chain from the Grandparent down to the ʼ473 Patent. Nor does Petitioner dispute that the ʼ473 Patent satisfies the other requirements of § 120. Instead, Petitioner’s argument hinges on the contention that because the Parent Application includes only claims that are directed to new matter, the Parent Application is not “similarly entitled to the benefit of the filing date of the” Grandparent Application, even though there is no break in the chain of disclosure from the Grandparent Application to the ʼ473 Patent.7 See Pet. 28–32. Petitioner argues that this is an issue of first impression and that “[t]here is no caselaw because this case presents a unique fact pattern.” Pet. Reply 25. Thus, Petitioner relies only on the language of 35 U.S.C. § 120 itself as support for its assertion that the Parent Application does not benefit from the filing date of the Grandparent Application. But we find no indication in § 120 that the Parent Application breaks the chain simply because the inventors chose to include claims directed only to new matter in the Parent Application, rather than including some claims directed to the

7 Petitioner does not dispute that the ’473 patent, as well as the Parent, incorporate by reference the prior applications in the chain up to the Grandparent Application. Pet. 29; Ex. 1001, 1:5–13.

16 IPR2020-00323 Patent 9,503,473 B1 disclosure of the Grandparent Application—an application that was properly incorporated by reference in the Parent Application. Petitioner has not demonstrated that there has been a break in the priority chain as to claims directed only to the common subject matter continuously disclosed by the applications in the chain. Because we find that the priority chain for the ʼ473 Patent properly continues back to the Grandparent Application through the Parent Application under § 120, the claims of the ʼ473 Patent directed only at the common subject matter benefit from the filing date of the Grandparent Application. Transco Prods., Inc. v. Performance Contracting, Inc., 38 F.3d 551, 557 n.6 (Fed. Cir. 1994) (“a continuing application is entitled to rely on the filing date of an earlier application only with respect to subject matter common to both applications”). 3. Conclusion In summary, we find that the invention of the ʼ473 Patent is sufficiently disclosed in the Grandparent Application and in all other descendant applications in the chain down to the ʼ473 Patent and that the other requirements of § 120 are met. This entitles the claims of the ’473 Patent to the benefit of the effective filing date of the Grandparent Application. For these reasons and based on the arguments and evidence presented in the Petition, we determine that Petitioner has not shown by a preponderance of the evidence that the ’445 Patent is prior art, and thus has not shown that the ʼ445 Patent anticipates the ʼ473 Patent.

17 IPR2020-00323 Patent 9,503,473 B1 D. Grounds under § 103(a) Petitioner contends claims 1–6, 8–17, and 19–29 would have been obvious in view of Waterson and Ross; and claims 1–6, 8–17, and 19–29 would have been obvious in view of Waterson, Ross, and Geon. A claim is unpatentable under § 103(a) if the differences between the claimed subject matter and the prior art are “such that the subject matter as a whole would have been obvious at the time the invention was made to a person having ordinary skill in the art to which said subject matter pertains.” KSR Int’l Co. v. Teleflex Inc., 550 U.S. 398, 406 (2007). The question of obviousness is resolved on the basis of underlying factual determinations, including: (1) the scope and content of the prior art; (2) any differences between the claimed subject matter and the prior art; (3) the level of skill in the art; and (4) when present, objective evidence of nonobviousness, i.e., secondary considerations. Graham v. John Deere Co., 383 U.S. 1, 17–18 (1966). Additionally, the obviousness inquiry typically requires an analysis of “whether there was an apparent reason to combine the known elements in the fashion claimed by the patent at issue.” KSR, 550 U.S. at 418 (citing In re Kahn, 441 F.3d 977, 988 (Fed. Cir. 2006) (requiring “articulated reasoning with some rational underpinning to support the legal conclusion of obviousness”)); see In re Warsaw Orthopedic, Inc., 832 F.3d 1327, 1333 (Fed. Cir. 2016) (citing DyStar Textilfarben GmbH & Co. Deutschland KG v. C. H. Patrick Co., 464 F.3d 1356, 1360 (Fed. Cir. 2006)). As explained below, we determine based on the present record that Petitioner has shown by a preponderance of the evidence that each of claims 1–6, 8–17, and 19–29 is unpatentable as obvious under 35 U.S.C. § 103(a).

18 IPR2020-00323 Patent 9,503,473 B1 1. Objective Indicia of Non-Obviousness The Graham factors instruct that, when present, we must consider— apart from what the prior art itself would have suggested—whether objective evidence of nonobviousness (i.e., secondary considerations) may lead to a conclusion that the challenged claims would not have been obvious. See, e.g., Stratoflex, Inc. v. Aeroquip Corp., 713 F.2d 1530, 1538–39 (Fed. Cir. 1983) (instructing that evidence of secondary considerations, when present, must always be considered in determining obviousness). Objective evidence of non-obviousness is only a part of the “totality of the evidence”; its mere existence does not control the conclusion of obviousness. See Richardson- Vicks Inc. v. Upjohn Co., 122 F.3d 1476, 1483 (Fed. Cir. 1997). Objective evidence of nonobviousness “may often be the most probative and cogent evidence in the record” and “may often establish that an invention appearing to have been obvious in light of the prior art was not.” Transocean Offshore Deepwater Drilling, Inc. v. Maersk Drilling USA, Inc., 699 F.3d 1340, 1349 (Fed. Cir. 2012). Objective evidence of nonobviousness may include evidence of commercial success, licensing, copying, praise by others, long- felt but unresolved need, and failure by others. Graham, 383 U.S. at 17–18. Objective evidence of nonobviousness “is only relevant to the obviousness inquiry ‘if there is a nexus between the claimed invention and the [objective indicia of nonobviousness].’” In re Affinity Labs of Tex., LLC, 856 F.3d 883, 901 (Fed. Cir. 2017) (quoting Ormco Corp. v. Align Tech., Inc., 463 F.3d 1299, 1312 (2006)). “[T]here is no nexus unless the evidence presented is ‘reasonably commensurate with the scope of the claims.’” In re Affinity Labs of Tex., 856 F.3d at 901 (citing Rambus Inc. v. Rea, 731 F.3d 1248, 1257 (Fed. Cir. 2013) (quoting In re Kao, 639 F.3d 1057, 1068 (Fed. Cir. 2011)).

19 IPR2020-00323 Patent 9,503,473 B1 Patent Owner argues “[t]he non-obviousness of the challenged claims is further supported by evidence of secondary considerations, including a long felt but unresolved need for the claimed invention, the failure of others, and commercial success.” PO Resp. 58. a) Long-Felt Need and Failure of Others According to Patent Owner, in early 2007, its predecessor company, Turiss, identified a rapid growth in criminal theft of user credentials caused by a new type of known as form grabbing key loggers. PO Resp. 59–60. This type of malware started being used more frequently with the release of “,” a form grabbing software. PO Resp. 60. Around February of 2007, Mr. Liske, the inventor of the ʼ473 Patent, devised a method to defeat Zeus, and other form grabbing malware, which did not require detecting the malware. PO Resp. 61 (citing Ex. 2044 ¶¶ 26–30). After filing a patent application, Turiss began marketing Mr. Liske’s anti- form grabbing solution to financial institutions. PO Resp. 61–62 (citing Ex. 2044 ¶¶ 33–34). One major bank with which Turiss met was previously unaware of the rise of form grabbing key loggers and needed to be educated about the problem. PO Resp. 62 (citing Ex. 2044 ¶¶ 33–34). After understanding the need, this bank became a customer and continues to be one. PO Resp. 62 (citing Ex. 2044 ¶¶ 33–34). In order to show a long-felt but unmet need for the claimed invention, the objective evidence must show that the need was a persistent one that was recognized by those of ordinary skill in the art at the time of the invention. In re Gershon, 372 F.2d 535, 538 (CCPA 1967). We find that Patent Owner does not present sufficient evidence showing that the need for protection against form grabbing malware was persistent, long-felt, or recognized generally by persons of ordinary skill in the relevant art. Mr. Reschke, the

20 IPR2020-00323 Patent 9,503,473 B1 Head of Threat Intelligence and Operations Manager for Trusted Knight, testifies that Turiss monitored and surveilled cybercriminals from 2006 onwards, and that in early 2007, it identified a rapid growth in criminal theft and sale of user credentials. Ex. 2044 ¶¶ 3, 9–12. At that time, those in the industry attributed the proliferation of stolen user credentials to , rather than form-grabbing malware. Ex. 2044 ¶ 12. In addition, a form- grabbing malware named “Zeus” was released in 2006 and drew the attention of the cyber-security industry in 2007. Ex. 2044 ¶ 17. Around February of 2007, Mr. Liske had devised a method to defeat Zeus, one that did not require detection of malware. Ex. 2044 ¶¶ 27–30. In April of 2008, Mr. Liske filed a patent application that eventually issued as the ʼ473 Patent. Ex. 2044 ¶ 32. Based on this evidence, the earliest recognition of the problem caused by form-grabbing malware appears to have occurred in early 2007, when Turiss identified a rapid growth in the theft of user credentials and when Zeus drew the attention of the cyber-security industry. Ex. 2044 ¶¶ 9–12, 17. About a year later, Mr. Liske applied for a patent on his invention. Ex. 2044 ¶ 32. No other evidence is presented to show that those skilled in the art, other than Mr. Liske or those at Turiss, recognized this need or made unsuccessful attempts at solving the problem. We do not find the period of about a year, between the time when Turiss recognized the problem and the filing of Mr. Liske’s patent application, to show a long and persistent need, particularly in the absence of evidence showing that others also had perceived this need and made unsuccessful attempts to solve the problem during this period. Ecolochem, Inc. v. S. Cal. Edison Co., 227 F.3d 1361, 1377 (Fed. Cir. 2000) (finding a one-year time between identification of the problem and patented solution was insufficient to show “a long-felt need”).

21 IPR2020-00323 Patent 9,503,473 B1 Accordingly, based on all the evidence and arguments presented, we do not weigh Patent Owner’s evidence of an alleged long-felt, unmet need or failure of others to meet this need as significantly supportive of Patent Owner’s position in the obviousness analysis. b) Commercial Success Patent Owner further argues that “[c]ommercial success is evidenced by IBM’s widespread adoption of the claimed invention.” PO Resp. 62. According to Patent Owner, in April 2008, an Israeli company named Trusteer purported to have a product named Rapport that defeated form grabbing software but that, at least in 2008, Rapport did not protect against Zeus. PO Resp. 62 (citing Ex. 2044 ¶¶ 35–37). By 2010, however, Trusteer “apparently used the Trusted Knight invention to develop and make a new version of Rapport.” PO Resp. 62. Trusteer was acquired by IBM in 2014. PO Resp. 62. According to Patent Owner, Trusteer estimated that around fifty banks were using its technology, and that IBM now claims that Rapport serves hundreds of customers and is running on millions of PCs. PO Resp. 62 (citing Ex. 2044 ¶¶ 35–37). As an initial matter, “[f]or objective evidence to be accorded substantial weight, its proponent must establish a nexus between the evidence and the merits of the claimed invention.” In re GPAC Inc., 57 F.3d 1573, 1580 (Fed. Cir. 1995). A patentee “cannot demonstrate commercial success, for purposes of countering the challenge of obviousness, unless it can show that the commercial success of the product results from the claimed invention.” J.T. Eaton & Co. v. Atl. Paste & Glue Co., 106 F.3d 1563, 1571 (Fed. Cir. 1997). Put another way, “objective evidence of non- obviousness must be commensurate in scope with the claims which the evidence is offered to support.” Asyst Techs., Inc. v. Emtrak, Inc., 544 F.3d

22 IPR2020-00323 Patent 9,503,473 B1 1310, 1316 (Fed. Cir. 2008) (quoting In re Grasselli, 713 F.2d 731, 743 (Fed. Cir. 1983)). “[T]here is a presumption of nexus for objective considerations when the patentee shows that the asserted objective evidence is tied to a specific product and that product ‘is the invention disclosed and claimed in the patent.’” WBIP, LLC v. Kohler Co., 829 F.3d 1317, 1329 (Fed. Cir. 2016). Even when a presumption is inappropriate, the patent owner is still afforded an opportunity to prove nexus by showing that the evidence of commercial success is the “direct result of the unique characteristics of the claimed invention.” In re Huang, 100 F.3d 135, 140 (Fed. Cir. 1996). Here, other than to state that Trusteer “apparently used the Trusted Knight invention to develop and make a new version of Rapport” (PO Resp. 62), Patent Owner provides no evidence that Rapport “embodies the claimed features and is coextensive with them.” Brown & Williamson Tobacco Corp. v. Philip Morris Inc., 229 F.3d 1120, 1130 (Fed. Cir. 2000). The record here is devoid of any comparison of Rapport to the claimed features or any evidence that the alleged success of Rapport was driven by any novel aspect of the claimed invention. J.T. Eaton, 106 F.3d at 1571 (“[C]ommercial success of the product must be due to the merits of the claimed invention beyond what was readily available in the prior art.”) Moreoever, there is no evidence in the record of the relevant market or of relative market of Rapport as compared to other products. Gross sales figures do not show commercial success absent evidence as to market share. Cable Electric Products, Inc. v. Genmark, Inc., 770 F.2d 1015, 1026– 27 (Fed. Cir. 1985); In re Applied Materials, Inc., 692 F.3d 1289, 1300 (Fed. Cir. 2012) (“[T]he number of units sold without evidence of the market share is only weak evidence of commercial success.”). Without such

23 IPR2020-00323 Patent 9,503,473 B1 evidence we lack the appropriate context to evaluate the significance of Patent Owner’s bare assertions that “around 50 banks were using [Trusteer’s] technology” or that Rapport now serves hundreds of customers. PO Resp. 62. Accordingly, based on the evidence and arguments presented, we do not weigh Patent Owner’s evidence of alleged commercial success as significantly supportive of Patent Owner’s position in the obviousness analysis. 2. Obviousness over Waterson and Ross a) Overview of Waterson Waterson relates to a method and system for securing user entry data against keyboard login attempts. Ex. 1009, Abstract. The method/system masks user input data, either insulating the data from vulnerable parts of the operating system or application, or contaminating user data that is supplied to those vulnerable parts of the operating system or application. Ex. 1009, Abstract. The method/system securely stores data, either intercepted prior to reaching the vulnerable parts of the operating system or extracted from the contaminated data, and securely provides the data directly to a verified application. Ex. 1009, Abstract. Figure 6 of Waterson is reproduced below:

24 IPR2020-00323 Patent 9,503,473 B1

Figure 6 depicts a representation of the different software levels at which keystroke loggers may operate. Ex. 1009, 4:6–7. The user enters sensitive information on a keyboard (or key-entry device) 601 which is connected to a computer (or other hardware) 608. Ex. 1009, 4:58–60. The operating system of the computer has a keyboard driver and buffer 602 that interfaces between the hardware computer keyboard and any software. Ex. 1009, 4:61–63. A kernel level program 603 harvests keystroke information directly from the keyboard driver. Ex. 1009, 4:63–65. Shell level software 604 typically interfaces between the keyboard driver of the operating system and other higher level software applications. Ex. 1009, 4:66–5:1. Shell level keystroke logging software 605 harvests information straight from the operating system keyboard buffer. Ex. 1009, 5:3–5. Application level keystroke logging software 607 further harvests keystroke information from an application 606. Ex. 1009, 5:6–8. Figure 5 of Waterson is reproduced below:

25 IPR2020-00323 Patent 9,503,473 B1

Figure 5 depicts the protection of sensitive keystroke data. Ex. 1009, 4:3–5. The user enters the data via keyboard 501, where the data from keyboard 501 enters keyboard buffer 502. Ex. 1009, 10:39–40. Unknown to the user of keyboard 501, a keyboard logger 505 is present. Ex. 1009, 10:40–42. The application 503 launches a data entry window 504 to gather data from the user, where the data entry window 504 includes a protected entry field 507. Ex. 1009, 10:42–45. When the entry field 507 is initialized, application 503 requests a token. Ex. 1009, 10:45–46. The entry field 507 generates a token and returns it to application 503. Ex. 1009, 10:46–48. When the entry field 507 gains focus, indicating that the user of keyboard 501 intends to enter data into the entry field 507, the entry field begins to stream random data 511 to the keyboard buffer 502. Ex. 1009, 10:49–52. The user of keyboard 501 enters data at keyboard 501 and this clean data 509 enters keyboard buffer 502. Ex. 1009, 10:52–54. The

26 IPR2020-00323 Patent 9,503,473 B1 content of the keyboard buffer 502 is read by entry field 507 and may also be read by a keystroke logger 505. Ex. 1009, 10:54–56. The entry field 507 receives contaminated data 513, and the keystroke logger 505 also receives contaminated data 512. Ex. 1009, 10:56–57. The entry field 507 generates and stores the random data stream 511 which contaminates the clean data 509. Ex. 1009, 10:58–59. The entry field 507 removes the random data from the contaminated data 513 that it receives from the buffer to leave clean data. Ex. 1009, 10:59–62. Once the focus leaves the secured entry field 507, the entry field 507 stops streaming data to the keyboard buffer. Ex. 1009, 10:62–63. The application 503 requests the entered data by sending a request accompanied by the token to the entry field 507. Ex. 1009, 10:63–65. Clean data 514 is supplied directly to the application 503 in response to the authenticated request. Ex. 1009, 10:65–67. False data 515 is supplied to the data entry window 504. Ex. 1009, 10:67–11:1. The keystroke logger 505 extracts the false data from the data entry window as indicated by arrow 510. Ex. 1009, 11:1–3. Accordingly, the secured entry field 507 defeats two of the methods by which a keystroke logger 505 attempts to extract sensitive information entered by the user of keyboard 501 while still allowing a user to supply the sensitive information the application 503 in the usual manner. Ex. 1009, 11:4–9. b) Overview of Ross Ross relates to a browser extension, PwdHash, that produces a different password for each site and defends against password phishing and other attacks. Ex. 1010, Abstract. The PwdHash browser extension applies a cryptographic hash function to a combination of the plaintext password entered by the user, data associated with the web site, and (optionally) a

27 IPR2020-00323 Patent 9,503,473 B1 private salt (i.e. random data used as input for a function that hashes data) stored on the client machine, such that the theft of the password received at one site will not yield a password that is useful at another site. Ex. 1010, Abstract. Upon detection of either a “password-prefix” (i.e., sequence of printable characters) or a “password-key” (i.e., dedicated keyboard key), the PwdHash browser extension replaces the user-entered password with a hashed version of the password. Ex. 1010, at 4, 6. The hashing can occur at one of two times: (1) replacing the contents of the field with the hashed password when focus leaves the field; or (2) trapping the form submission event and replacing the contents of all password fields with the appropriate hashed passwords. Ex. 1010, at 4. c) Whether Ross is Analogous Art A reference qualifies as prior art for an obviousness determination under 35 U.S.C. § 103 only when it is analogous to the claimed invention. In re Klein, 647 F.3d 1343, 1348 (Fed. Cir. 2011); see also Innovention Toys, LLC, v. MGA Entertainment, Inc., 637 F.3d 1314, 1321 (Fed. Cir. 2011); In re Bigio, 381 F.3d 1320, 1325 (Fed. Cir. 2004); In re Clay, 966 F.2d 656, 658 (Fed. Cir. 1992). Two separate tests define the scope of analogous prior art: (1) whether the art is from the same field of endeavor, regardless of the problem addressed or, (2) if the reference is not within the field of the inventor’s endeavor, whether the reference still is reasonably pertinent to the particular problem with which the inventor is involved. Klein, 647 F.3d at 1348. Petitioner argues that Waterson and Ross are in the same field of endeavor as the ʼ473 Patent, as all three references teach methods for protecting from malware that grab keystrokes out of forms on a webpage. Pet. 61. Petitioner argues that Waterson teaches anti-keylogging software to

28 IPR2020-00323 Patent 9,503,473 B1 protect an entry field and Ross teaches a solution for hashing passwords to prevent their capture by keylogging software. Pet. 61 (citing Ex. 1009, 10:38–64; Ex. 1010, 4). In addition, Petitioner emphasizes that Ross’s technique of trapping form submission events is the same solution taught by the ʼ473 Patent to defeat form grabbing keylogging malware. Pet. 61–62 (citing Ex. 1001, 5:48–60). Patent Owner argues that Ross is not in the same field of endeavor as the ʼ473 Patent because the ʼ473 Patent is directed to software for combating form grabbing malware, whereas Ross is directed to combating phishing scams. PO Resp. 43. Patent Owner argues that Ross emphasizes that it “does not ‘defend[] against and keyloggers already installed on the user’s machine.’” PO Resp. 43 (quoting Ex. 1010, 6) (alteration in original). Patent Owner also contends that Ross is only concerned with attacks from malicious phishing sites and that its browser extension improves password security, but ‘does not protect user passwords from spyware, keyloggers, and other software that is installed on the local machine.’” PO Resp. 43–44 (quoting Ex. 1010, 12). Patent Owner also argues that, for largely the same reasons, Ross fails the reasonably pertinent test. PO Resp. 44. Namely, Patent Owner argues that because Ross “admits that ‘it does not protect user passwords from [] keyloggers’” one of ordinary skill would not have naturally looked to Ross to incorporate its teachings into the anti-key logging software described in Waterson. Petitioner disagrees with Patent Owner’s argument that Ross does not defend against key loggers, arguing that Ross teaches that its “PwdHash solution is effective against ‘JavaScript keyloggers’” and that when

29 IPR2020-00323 Patent 9,503,473 B1 implemented at the kernel level, it can “defeat keyloggers that ‘listen to keyboard events.’” Pet. Reply 20. We first determine whether Ross is in the same field of endeavor as the ʼ473 Patent. To determine the applicable field of endeavor, the factfinder must consider “explanations of the invention’s subject matter in the patent application, including the embodiments, function, and structure of the claimed invention.” Bigio, 381 F.3d at 1325. In addition the factfinder must consider each reference’s disclosure in view of the “reality of the circumstances” and “weigh those circumstances from the vantage point of the common sense likely to be exerted by one of ordinary skill in the art in assessing the scope of the endeavor.” Bigio, 381 F.3d at 1326. The ʼ473 Patent states that it “relates to systems and methods for protection against the operation of malware commonly used in identity-theft and cyber-fraud.” Ex. 1001, 1:18–20. The ʼ473 Patent discusses two types of malware in particular, hook-based key loggers and form grabbing key loggers. Ex. 1001, 1:65–2:64. Although the ʼ473 Patent focuses “[i]n particular, but not by way of limitation, . . . [on] systems and methods for preventing key logger malware that utilizes form grabbing techniques to steal financial and identity information from users’ browsers” (Ex. 1001, 1:20–24), it discloses “[e]mbodiments . . . [that] also relate to an apparatus, system, and method to prevent hook-based keyloggers from stealing information” (Ex. 1001, 4:5–8). As such, we determine that the field of endeavor of the ʼ473 Patent is protection against key logging malware. Ross is directed to “a browser extension, PwdHash, that transparently produces a different password for each site, improving web password security and defending against password phishing and other attacks.” Ex. 1010, 1. Ross explains that in a phishing scam, an attacker sets up a website

30 IPR2020-00323 Patent 9,503,473 B1 that masquerades as a legitimate site and obtains the user’s cleartext password for the legitimate site. Ex. 1010, 1. One type of phishing attack addressed by Ross is a JavaScript attack that listens to keyboard events sent to the password field and records those keys. Ex. 1010, 2–3. Ross, explains that with its solution, these “JavaScript keyloggers . . . cannot steal the cleartext password.” Ex. 1010, 4. Ross goes on to explain, at various points, that while its solution can defend against scripts on remote web sites, it “does not protect user passwords from spyware, keyloggers, and other software that is installed on the local machine.” Ex. 1010, 12 (emphasis added); see also Ex. 1010, 6 (explaining that Ross does not defend against spyware or keyloggers already installed on the user’s machine, but then proposing a solution where its technique is implemented inside the OS kernel so that a hashed version of the password can be embedded directly into outgoing HTTP requests.) From this description it is clear Ross protects against key logging when that key logging occurs on a remote web site. Although Patent Owner focuses on sections of Ross explaining that its solution does not protect against key loggers, those sections clearly reference local key loggers installed on the machine itself, not key loggers in general. Moreover, even in the instance of locally installed key loggers, Ross proposes a solution in which its hashing technique is implemented at a kernel level. Ex. 1010, 6. Accordingly, we determine that one of ordinary skill in the art would have viewed both the ʼ473 Patent and Ross as directed at protecting against key logging malware, and thus to be in the same field of endeavor. Because we determine that Ross and the ʼ473 Patent are in the same field of endeavor we do not address whether Ross is reasonably pertinent to the particular problem with which the inventor is involved.

31 IPR2020-00323 Patent 9,503,473 B1 d) Combination of Waterson and Ross Petitioner incorporates Ross’s solution for trapping browser form submission events with Waterson’s anti-key logging software in order to protect against key loggers that grab keystrokes during form submission. Pet. 59 (citing Ex. 1013 ¶¶ 100–108). According to Petitioner, in Waterson, the anti-key logging software is activated when a form is in focus. Pet. 67. In the combined system, however, the anti-key logging software is activated when a form submission initiation call event, such as the BeforeNavigate2 event described in Ross, is detected. Pet. 67 (citing Ex. 1010, 4; Ex. 1013 ¶¶ 116–117). Petitioner argues that the motivation to have combined Ross’s form submission event solution with Waterson’s anti-keylogging software is to provide an additional level of protection against form grabbing key logging software that Waterson alone may not provide. Pet. 59 (citing Ex. 1013 ¶¶ 100–108). Specifically, Petitioner argues “Waterson teaches that the ‘protected entry field’ ‘supplies the unmasked user input data to the underlying application that is authorized to receive it,’ leaving open the possibility that the data could still be captured when a form is submitted.” Pet. 64 (citing Ex. 1009, 8:48–56). According to Petitioner, Ross fills this gap by intercepting the form submission event with a BeforeNavigate2 handler, cancelling the original Navigate2 event and firing a new, modified one. Pet. 64 (citing Ex. 1010, 11). In the combination, Ross’s browser helper object solution would provide protection against browser-level, form grabbing key loggers, which do not attempt to log keys at the kernel level, and thus, would not be prevented by the kernel-level solution described in Waterson. Pet. 64. Petitioner argues that Waterson’s “anti-keylogging software can ‘defeat keystroke loggers that attempt to harvest keystroke

32 IPR2020-00323 Patent 9,503,473 B1 information at a kernel level.’” Pet. 61 (quoting Ex. 1009, 6:18–22). Ross, on the other hand, recognizes that while its proposed solution, the PwdHash browser extension, cannot ‘defend[] against spyware and keyloggers already installed on the user’s machine’ this problem can be solved ‘implement[ing] the password-prefix/password-key mechanism inside the OS kernel.’” Pet. 61 (citing Ex. 1010, 4, 6) (emphasis removed, alterations in original). Petitioner argues that “[c]ombining Waterson and Ross would address a core principle in referred to as ‘complete mediation.’” Pet. 59 (citing Ex. 1013 ¶ 102). Complete mediation, according to Petitioner, is the principle that dictates that every use of a particular security- sensitive interface or device must be protected. Pet. 59. Patent Owner argues that keyboards are one such device, and that “implementing Ross’s form trap mechanism to bring additional protection to the browser represents an important step toward providing complete mediation for the user and computer environment.” Pet. 60. Additionally, as argued by Petitioner, both Waterson and Ross address the same problem: defeating malicious software that can steal a user’s confidential data. Pet. 60. Petitioner argues that, like the ’473 Patent, both Waterson and Ross teach methods for protecting from malware that grab keystrokes out of forms on a web page. Id. at 61. As further argued by Petitioner, a person of ordinary skill in the art would have had a reasonable expectation of success in incorporating the form submission event trap disclosed in Ross into the anti-keylogging software disclosed in Waterson at least because Waterson describes that its software can be incorporated within a local application such as Internet browsing software. Id. at 62. Both systems, according to Petitioner, detect forms in a webpage and take steps to protect keystrokes entered into those forms from malware operating

33 IPR2020-00323 Patent 9,503,473 B1 at the operating system and kernel levels. Pet. 63 (citing Ex. 1009, 5:10–19, 5:49–50; Ex. 1010, 6). Thus, Petitioner argues, a person of ordinary skill in the art would merely have had to incorporate Ross’s form submission trap inside the kernel, as described in Waterson, and that this would have been a simple modification since Waterson’s anti-key logging software already operates at the kernel level. Pet. 63 (citing Ex. 1013 ¶ 107). Patent Owner argues that one of ordinary skill would not have been motivated to combine Waterson and Ross for several reasons. PO Resp. 34– 42. For example, Patent Owner argues that Ross teaches away from combining the two references (PO Resp. 36–38), that any motivation to combine is the result of improper hindsight (PO Resp. 38–39), that combining Waterson and Ross would render Waterson inoperable for its intended purpose (PO Resp. 39–42), that Petitioner has failed to demonstrate a reasonable expectation of success in incorporating Ross’s form submission trap event into Waterson (PO Resp. 45–46), and that Petitioner has not set forth the combination with the requisite particularity (PO Resp. 46–47). We analyze these arguments further below. A motivation to combine may be found “explicitly or implicitly in market forces; design incentives; the ‘interrelated teachings of multiple patents’; ‘any need or problem known in the field of endeavor at the time of invention and addressed by the patent’; and the background knowledge, creativity, and common sense of the person of ordinary skill.” ZUP, LLC v. Nash Mfg., Inc., 896 F.3d 1365, 1371 (Fed. Cir. 2018) (quoting Plantronics, Inc. v. Aliph, Inc., 724 F.3d 1343, 1354 (Fed. Cir. 2013)). We have carefully considered all arguments and supporting evidence regarding the rationale for combining the teachings of Waterson and Ross. We are persuaded by Petitioner that a person of ordinary skill would have been motivated to

34 IPR2020-00323 Patent 9,503,473 B1 combine Waterson’s anti-keylogging software with Ross’s form submission event solution to provide an additional level of protection against form grabbing key logging software that Waterson alone does not provide. Dr. McDaniel provides credible testimony that incorporating Ross’s browser- level solution into Waterson would provide added benefits for browser applications that Ross recognizes are a common target of keyloggers. Ex. 1013 ¶ 100. Dr. McDaniel also testifies that the principle of complete mediation, whereby every use of a security-sensitive interface (such as a keyboard) is protected, would motivate one of ordinary skill in the art to use Ross’s form trap submission mechanism to bring protection to the browser in Waterson’s software. Ex. 1013 ¶ 102. We, therefore, disagree with Patent Owner’s argument that Petitioner’s combination is “tainted” by improper hindsight. PO Resp. 38. Instead, Petitioner has articulated a reason with rational underpinning for combining the two references. We also do not agree with Patent Owner’s argument that Petitioner has failed to show a reasonable expectation of success in combining Ross with Waterson or that Petitioner has not explained the combination with the requisite particularity. Dr. McDaniel provides credible testimony that Waterson describes that its software can be incorporated within a local application such as an Internet browser like a plugin, which is how the Ross’s software is implemented as well. Ex. 1013 ¶ 106 (citing Ex. 1009, 11:45–48; Ex. 1010, 11). Furthermore, Dr. McDaniel credibly testifies that Ross suggests to incorporate its solution inside the operating system kernel, which is precisely the combination proposed by Petitioner. Ex. 1013 ¶ 107 (citing Ex. 1010, 6). The suggestions in Waterson to incorporate its software in an Internet browser as a plugin, and the suggestion in Ross to implement its software in the OS kernel, show that one of skill in the art would have

35 IPR2020-00323 Patent 9,503,473 B1 had a reasonable expectation of success in making these modifications. Accordingly, we determine that, not only has Petitioner explained the combination with particularity, but Petitioner has provided sufficient evidence showing that combining Waterson and Ross would have been within the level of skill of the ordinary artisan. We do not agree with Patent Owner’s argument that Ross teaches away from combining with Waterson. PO Resp. 36–38. Patent Owner argues “Ross particularly notes that it cannot ‘defend[] against spyware and keyloggers’ . . . and recognizes that any such functionality is for ‘promising direction for future research.’” PO Resp. 37 (citing Ex. 1010, 6; Ex. 2043 ¶¶ 17–27) (first alteration in original). Thus, according to Patent Owner, “Ross criticizes, discredits and discourages use of its ‘browser extension, PwdHash’ to combat keylogging malware.” PO Resp. 37. As explained above in our discussion of whether Ross is analogous art, Ross makes clear that its solution does not “defend[] against spyware and key loggers already installed on the user’s machine” not that it does not defend against key loggers in general. Ex. 1010, 6 (emphasis added). On the contrary, Ross states that its solution addresses JavaScript keyboard monitoring and JavaScript key loggers which are located on phishing websites rather than locally on the user’s machine. Ex. 1010, 3, 4. Ross recognizes this deficiency and proposes a solution—installing its software in the OS kernel. Ex. 1010, 6. Thus, rather than discouraging the combination with Waterson, Ross instead explicitly suggests implementing its solution in the kernel, just as proposed by Petitioner in its combination of Waterson and Ross, and encourages the reader to do so by characterizing this solution as “a promising direction for future research.” Ex. 1010, 6.

36 IPR2020-00323 Patent 9,503,473 B1 We also disagree with Patent Owner’s argument that combining Waterson and Ross would render Waterson inoperable for its intended purpose. PO Resp. 39. Here, Patent Owner argues that Ross relates to password managers, not key loggers and that combining Ross with Waterson would frustrate Waterson’s intended purpose of combatting malware. PO Resp. 40 (citing Ex. 2043 ¶¶ 6–10). As explained above, we find that Ross does, in fact, relate to addressing key logging on phishing websites and therefore we disagree with this argument. Ex. 1010, 2–4. Patent Owner further argues that combining Ross with Waterson would incorporate not just Ross’s form submission event into Waterson, as Petitioner proposes, but also Ross’s browser extension PwdHash, resulting in incoming data (such as passwords) being hashed and thereby preventing Waterson from receiving any clean data. PO Resp. 21, 42. We disagree. Petitioner does not incorporate Ross’s hashing function in the combination of Waterson and Ross. Petitioner is correct that the obviousness analysis does not require bodily incorporation of one reference with another. In re Keller, 642 F.2d 413, 425 (CCPA 1981). One of ordinary skill would, in light of the teachings of Waterson and Ross, be able to incorporate Ross’s solution for trapping browser form submission events with Waterson’s software without also necessarily incorporating the password hashing function of Ross. e) Analysis of Limitations of Claims 1 and 11 Petitioner presents a detailed explanation of how the combination of Waterson and Ross teaches each of the limitations of claim 1. Pet. 64–70. Petitioner relies on these same contentions for claim 11. Pet. 74–75. Patent Owner presents the same arguments for claim 1 and claim 11. PO Resp. 13. Because Petitioner and Patent Owner rely on the same arguments for claims

37 IPR2020-00323 Patent 9,503,473 B1 1 and 11, our discussion below focuses primarily on the limitations of claim 1. Further, our discussion below incorporates our consideration of Patent Owner’s objective evidence of nonobviousness as discussed above. Specifically, when discussing motivation or suggestion to combine, we have considered the entirety of the evidence, both for and against obviousness. (1) [1pre] “A method for preventing software key logging executable by a microprocessor, comprising” Petitioner argues Waterson teaches software that can prevent keystroke logging software. Pet. 64 (citing Ex. 1009, 1:10–14; Ex. 1013 ¶¶ 109–111). Petitioner argues that Waterson teaches that its software runs on a computer system and that computer systems execute software using microprocessors. Pet. 64 (citing Ex. 1013 ¶¶ 109–110). Patent Owner does not explicitly contest Petitioner’s arguments regarding the preamble of claim 1. See generally PO Resp. Generally, a preamble does not limit a claim. Allen Eng’g Corp. v. Bartell Indus., Inc., 299 F.3d 1336, 1346 (Fed. Cir. 2002). Here, we need not decide whether the preamble limits the claims because Petitioner establishes by a preponderance of the evidence that Waterson and Ross teach the preamble of claim 1. Waterson discloses “software that can prevent keystroke logging software from accessing or identifying keystrokes and mouse clicks entered by a user on a computer system.” Ex. 1009, 1:10–14. Dr. McDaniel provides credible testimony that computer systems “execute software using microprocessors, such as CPUs.” Accordingly, we are persuaded that the combination of Waterson and Ross teaches the preamble of claim 1.

38 IPR2020-00323 Patent 9,503,473 B1 (2) [1a] “installing and maintaining an anti-key lodger [sic] at a most privileged access level for browser events in an Application Programming Interface (API) stack.” Petitioner argues Waterson teaches anti-keylogging software that can “be installed ‘as a stand-alone software application,’ or ‘as part of the web browser process.’” Pet. 64–65 (citing Ex. 1009, 1:10–14, 11:14–17). According to Petitioner, Waterson’s software periodically verifies that it is at the top of the driver priority list, and if not, replaces itself at the top of the driver priority list by requesting the user to reboot the system. Pet. 65 (citing Ex. 1009, 6:32–38). Petitioner argues that by maintaining itself at the top of the driver priority list, it is also at the top of the Windows API stack, which, according to Petitioner, the ʼ473 Patent recognizes as the most privileged access level. Pet. 65 (citing Ex. 1001, Figs. 1, 3; Ex. 1013 ¶¶ 112–113). Petitioner relies on Ross’s disclosure of “form submission events” such as Internet Explorer’s BeforeNavigate2, as teaching the claimed “browser events.” Pet. 66 (citing Ex. 1010, 4). Petitioner argues the combination of Waterson and Ross therefore teach limitation 1a. Patent Owner argues Waterson does not disclose an anti-key logger that maintains itself at a most privileged access level for two main reasons. PO Resp. 15–17. First, Patent Owner argues that Waterson’s software merely verifies that it is at the top of a driver priority list and does so based on detection of another program, which is in contrast to the claimed software which does not depend on detection of malware at all. PO Resp. 15 (citing Ex. 2043 ¶¶ 48–51). Second, Patent Owner argues that Waterson does not teach “maintaining” at a most privileged access level because Waterson has no “automatic mechanism” to thwart the malware unless the user steps in to reboot. PO Resp. 16. According to Patent Owner, if the user does not

39 IPR2020-00323 Patent 9,503,473 B1 reboot the system, the software is unable to get back on top of the driver priority list and thus is unable to maintain itself at a most privileged access level. PO Resp. 17. Patent Owner further argues that Waterson and Ross do not disclose “browser events.” PO Resp. 17–19. Specifically, Patent Owner argues that a form grabber would still be able to steal the password even with Ross’s BeforeNavigate2 event being incorporated into Waterson. PO Resp. 18 (citing Ex. 1010, 4, 11; Ex. 2043 ¶¶ 10, 40, 52). Patent Owner also raises several arguments that are directed to whether one of ordinary skill in the art would have been motivated to combine Waterson and Ross. PO Resp. 18– 19. We addressed these arguments above in our analysis of the motivation to combine Waterson and Ross. See supra Part II.D.2.c)–d). Patent Owner argues that Waterson’s “driver priority list” is not part of the Windows API stack. PO Resp. 19. Patent Owner also argues that Waterson only discloses an API stack for keyboard events, not for a web browser or browser events. PO Resp. 20 (citing Ex. 2043 ¶ 40). Patent Owner also emphasizes that Waterson does not disclose the browser API function “Httpsendrequest” for Microsoft Windows, which was disclosed in the ʼ473 Patent. PO Resp. 20 (citing Ex. 1001, 9:33–34). We find that Petitioner’s arguments are supported by the evidence cited. Waterson discloses that its software “can execute on a computer system . . . as part of a web browser process (such as a plug-in, extension, help object, or control of the browser).” Ex. 1009, 11:14–19. Waterson further discloses The software of the present invention is preferably designed to remain at the top of the driver priority list while it is activated and periodically verifies this is the case. If at any time another program gains priority over the keyboard driver filter the

40 IPR2020-00323 Patent 9,503,473 B1 software alerts the user via a pop-up window and prompts them to reboot the system. Upon reboot the software replaces itself at the top of the driver priority list. Ex. 1009, 6:32–38. Dr. McDaniel provides credible testimony that by being at the “top of the driver priority list” Waterson’s software is installed at the zero-ring level, and therefore a most privileged access level. Ex. 1013 ¶ 112 (citing Ex. 1001, Figs. 1, 3). Dr. McDaniel also credibly testifies that by periodically checking that the software is installed first in the driver priority list, and prompting the user to restart if it is not so that it can replace itself back at the top of the list, Waterson teaches maintaining the software at the most privileged access level. Ex. 1024 ¶ 17. Patent Owner’s argument would require not only that the process happen automatically, but that it would exclude any user action, such as rebooting the computer. As explained above in our analysis of the construction of the “maintain” term, this argument is not commensurate with the scope of the claim, which does not require the process to occur automatically and also does not preclude user actions as part of the process. We also disagree with Patent Owner that Waterson does not teach maintaining its software at a most privileged access level because it requires detecting another program. This argument is also not commensurate with the scope of claim 1, which does not require providing for protection against malware without detecting it. Indeed, neither claim 1, nor claim 11, include such a limitation. Regardless, as we explain in our analysis of claim 2 below, which does include a limitation requiring “the clearing of the confidential data [to be] performed without requiring detection of the key logging malware,” we disagree with this argument. See infra Part II.D.2.f).

41 IPR2020-00323 Patent 9,503,473 B1 We disagree with Patent Owner’s argument that the combination of Ross and Waterson do not teach the claimed “browser events.” Ross teaches trapping form submission events such as BeforeNavigate2 in Internet Explorer. Ex. 1010, 4. Dr. McDaniel provides credible testimony that the BeforeNavigate2 event in Internet Explorer is a “browser event” as claimed. Ex. 1013 ¶ 113. Finally, we disagree with Patent Owner that Waterson and Ross do not disclose “an Application Programming Interface (API) stack.” Figure 3 of the ʼ473 Patent, as modified by Petitioner, is reproduced below.

Pet. 66. According to the ʼ473 Patent, Figure 3 “shows block diagrams of the API stacks with and without keyloggers.” Ex. 1001, 4:60–62. Thus, the ʼ473 Patent itself indicates that hardware drivers are part of the API stack.

42 IPR2020-00323 Patent 9,503,473 B1 In light of the aforementioned disclosure, by explaining that its software is “designed to remain at the top of the driver priority list,” (Ex. 1009, 6:33) we are persuaded that Waterson teaches that its software is at the zero-ring level and is part of the API stack. Dr. McDaniel supports this conclusion, testifying that “[t]he ‘driver priority list’ is part of the Windows API stack and is, as the ’473 Patent recognizes, at the “top” of the Windows API stack.” Ex. 1013 ¶ 112. Patent Owner’s argument that Waterson only discloses an API stack for keyboard events, not for a web browser or browser events is not commensurate with the scope of claim 1. Claim 1 has been construed to require “an API stack, accessed by a web browser application, including web browser and operating system APIs.” Petitioner provides credible evidence that Waterson’s driver priority list is part of the Windows operating system API. Pet. Reply 6 (citing Ex. 1013 ¶ 112; Ex. 1024 ¶ 23). (3) [1.b] “detecting, by the anti-key logger, a browser form submission initiation call event associated with data inputs entered by a user, wherein the form submission initiation call event is an OnSubmit call event or a BeforeNavigate call event In Petitioner’s combination, Waterson’s anti-key logging software is activated when a form submission initiation call event is detected, such as a BeforeNavigate2 event in Internet Explorer taught by Ross. Pet. 67. Petitioner argues that BeforeNavigate2 is a “form submission initiation call event.” Pet. 67 (citing Ex. 1013 ¶ 116). Patent Owner presents several arguments why the combination of Waterson and Ross does not disclose this limitation. For example, Patent Owner argues that Ross does not combat key loggers (PO Resp. 21), that the incorporation of Ross with Waterson would prevent Waterson from supplying clean data because the passwords would be hashed (PO Resp. 21–

43 IPR2020-00323 Patent 9,503,473 B1 22), and that “complete mediation” is a generic motivation to combine (PO Resp. 22). These arguments, however, have been addressed above in our discussion regarding the motivation to combine Waterson and Ross. See supra Part II.D.2.c)–d). In addition, Patent Owner also argues that Ross does not disclose an anti-key logger and therefore, even if it discloses a form submission initiation call event, this event is not detected by an anti-key logger. PO Resp. 21. Petitioner’s arguments are supported by the cited evidence. Ross discloses that it replaces a user’s password with a hashed version of the password and that the hashing can take place when a form submission event such as BeforeNavigate2 in Internet Explorer occurs. Ex. 1010, 4. Petitioner takes Ross’s ability to trap a browser form submission event and incorporates it into Waterson’s anti-key logger software so that Waterson’s software is triggered when a form submission event takes place. Pet. 67 (citing Ex. 1010, 4; Ex. 1013 ¶¶ 116–117). Thus, the combination of Waterson with Ross would teach “detecting, by the anti-key logger, a browser form submission initiation call event associated with data inputs entered by a user, wherein the form submission initiation call event is an OnSubmit call event or a BeforeNavigate call event.” We do not agree with Patent Owner’s argument that because Ross does not disclose an anti-key logger, its detection of a form submission initiation call event (such as the BeforeNavigate2 event) does not teach that an anti-key logger performs the detection. This argument ignores the combination proposed by Petitioner, in which Ross’s functionality is incorporated with Waterson’s anti-key logger software. Thus, in the

44 IPR2020-00323 Patent 9,503,473 B1 combination, an anti-key logger (Waterson’s anti-key logger) would perform the detection of the form submission event. (4) [1.c] “submitting the data inputs to a designated entity” Petitioner relies on Waterson’s disclosure of “suppl[ying] the unmasked user input data to the underlying application that is authorised to receive it” as teaching the aforementioned limitation. Pet. 68 (quoting Ex. 1009, 8:48–52) (emphasis removed). Petitioner argues that this application may be a banking website and “[a]fter a user completes the required ‘entry of sensitive information,’ the unmasked user inputs are submitted to the designated entity—the bank.” Pet. 68 (citing Ex. 1009, 11:28–44; Ex. 1013 ¶¶ 118–119). Patent Owner argues that under the agreed upon construction, a “designated entity” is an “intended receiving entity over a network.” PO Resp. 23 (citing Ex. 2034, 2). In Waterson, however, references to a banking website does not teach submitting data over a network because, according to Patent Owner, Waterson’s computer is a non-network computer and Waterson’s software is limited to a local computer that works with a local application. PO Resp. 23–24 (citing Ex. 2043 ¶¶ 65–66). Petitioner responds that Waterson teaches that the underlying application to which data is input may be web browsers and banking websites and that this information is submitted over the Internet. Pet. Reply 9 n.4 (citing Ex. 1009, 11:28–44; Fig. 6) We find that Petitioner’s arguments are supported by the cited evidence. Waterson discloses that its software can execute as a stand-alone software application, but it also discloses that it can be part of the web- browser process. Ex. 1009, 11:14–19. The software may be loaded onto the computer from a webpage such as an “internet banking website.” Ex. 1009,

45 IPR2020-00323 Patent 9,503,473 B1 11:28–32. Finally, Waterson depicts in Figure 6 that it protects against text grabbing keystroke loggers that log keystrokes from a web browser. See supra Part II.D.2.a). These combined disclosures would have taught one of ordinary skill in the art that Waterson’s computer is not non-networked and that data input into a web browser would be submitted over the Internet. Dr. McDaniel provides credible testimony in this regard. Ex. 1013 ¶ 119; Ex. 1024 ¶¶ 34–37. For example, Dr. McDaniel testifies that one type of application that Waterson protects are web browsers and that “[i]t is well- known that web browser communicates to websites over the Internet.” Ex. 1024 ¶¶ 35–36. (5) [1.d] “clearing, by the anti-key logger, confidential data from the data inputs to protect against the threat of key logging malware capturing the confidential data.” Claim 1 recites “clearing, by the anti-key logger, confidential data from the data inputs to protect against the threat of key logging malware capturing the confidential data.” Ex. 1001, 13:14–16. Petitioner relies on Waterson as teaching that when a user starts to enter data into an entry field, its anti-key logging software sends a ‘random data stream [] which has contaminated the clean data’ to the keyboard buffer, which is then passed to the entry field.” Pet. 68–69 (citing Ex. 1009, 10:49–57; Ex. 1013 ¶ 121) (emphasis removed, alteration in original). In the underlying litigation, the claim term “clearing confidential data” was agreed upon to mean “removing the meaning from confidential data.” See Pet. 21. Using this construction, Petitioner argues that by contaminating the clean sensitive data with a random data stream, Waterson teaches “removing the meaning form confidential data” for the purpose of protecting against the threat of key

46 IPR2020-00323 Patent 9,503,473 B1 logging malware capturing the confidential data. Pet. 69–70 (citing Ex. 1013 ¶¶ 120–122). Patent Owner argues that in the ʼ473 Patent, clearing of data happens after the web form data has been submitted to a designated entity, whereas in Waterson, clearing happens before data submission while the data is on its way to the web form. PO Resp. 25 (citing Ex. 1001, 6:10–17, 6:19–22, 12:2–4). Patent Owner further argues that Waterson’s mixing of random data with clean data does not correspond to “clearing” data because the clearing of data in the ʼ473 Patent is performed in a browser’s input or password fields, whereas in Waterson, the contaminated data is transmitted from the keyboard buffer to the entry field. PO Resp. 25–26 (citing Ex. 1001, 4:31– 32, 5:41–64, 13:31–32, 14:14–15, Fig. 2). Furthermore, Patent Owner argues that the data in Waterson is not cleared because the random data is tracked by the entry field in Waterson and therefore the random data can later be removed leaving the clean data that is provided to the application to be exposed in the memory of the application. PO Resp. 26. Patent Owner argues that “malware in the memory of the application can equally well access the random data stream, remove the random data, and thereby de-obfuscate the data inputs.” PO Resp. 26–27. Patent Owner argues that masking data does not teach clearing the data because “masking only temporarily obfuscates clean data.” PO Resp. 26. We are persuaded by Petitioner’s arguments that by masking confidential data with a random data stream, Waterson teaches “clearing, by the anti-key logger, confidential data from the data inputs to protect against the threat of key logging malware capturing the confidential data.” Dr.

47 IPR2020-00323 Patent 9,503,473 B1 McDaniel provides credible testimony that by contaminating keyboard inputs Waterson seeks to protect sensitive data such as bank account numbers, passwords, and logon information. Ex. 1013 ¶ 123. Dr. McDaniel testifies that by contaminating clean data Waterson teaches removing the meaning from confidential data. Ex. 1013 ¶ 122. We agree that when encountering a password, user information, or banking information, for example, which has been contaminated with random data, a key logger would not be able to discern the true meaning of the data. Thus, under the agreed upon construction of the term, which we apply here, Waterson’s masking of confidential data teaches the claimed “clearing” of confidential data. We disagree with Patent Owner’s argument that Waterson does not teach the claim limitation because in Waterson, clearing happens before data submission while the data is on its way to the web form. As an initial matter, claim 1 does not require clearing to occur in the web form. Regardless, the fact that in Waterson the clearing or contamination of data occurs while data is on its way to the web form does not mean that the data in the web form is not cleared. On the contrary, if the contamination occurs while the data is on its way to the web form, the data would already be contaminated when it arrived at the web form, and thus would have had its meaning removed in the web form. We disagree also with Patent Owner’s argument that because the contamination of data in Waterson is temporary, Waterson does not teach clearing of confidential data. Under the agreed upon construction, when Waterson contaminates the data inputs, it removes meaning from the data inputs and therefore clears the data, even if, later, the original data can be accessed by those authorized to access it.

48 IPR2020-00323 Patent 9,503,473 B1 (6) Conclusion For the reasons explained above, we determine that Petitioner has demonstrated by a preponderance of the evidence that the combination of Waterson and Ross teaches the limitations of claims 1 and 11 of the ʼ473 Patent. f) Claims 2 and 12 Claim 2 depends from claim 1 and recites “clearing of the confidential data is performed without requiring detection of the key logging malware.” Claim 12 depends from claim 11 and recites the same limitation. Petitioner relies on Waterson’s disclosure that it improves on conventional anti-key logging solutions by making it unnecessary to search for known key loggers in the user’s system or to search for files that match the signatures of known key loggers as evidence that Waterson clears confidential data without detection of key logging malware. Pet. 70 (citing Ex. 1009, 1:41–44, 4:12– 21, 11:45–53). Petitioner further argues that Waterson protects against key loggers whether the key loggers are present or not. Pet. 70–71 (citing Ex. 1009, 11:45–53; Ex. 1013 ¶¶ 124–125). Patent Owner argues, as it did for claim 1, that Waterson’s software is not maintained at a most privileged access level if another program gains priority. PO Resp. 28. When this happens, according to Patent Owner, Waterson’s software is no longer permanently activated and is no longer capable of clearing confidential data. PO Resp. 28 (citing Ex. 2043 ¶¶ 79– 82). We are persuaded by Petitioner’s argument that Waterson clears confidential data without detection of key logging malware. Dr. McDaniel credibly testifies that Waterson’s software is permanently activated and therefore protects against keylogging malware whether that malware is

49 IPR2020-00323 Patent 9,503,473 B1 present or not. Ex. 1013 ¶ 125. In other words, when no malware is present, Waterson would still clear confidential data and would do so, without detection of key logging malware. Moreover, even if malware were present but had not gained priority over Waterson’s software, the evidence shows that Waterson would clear confidential data and protect against the malware without requiring detection of the malware. Petitioner and Patent Owner rely on the same arguments for both claims 2 and 12. See Pet. 74–76; PO Resp. 33. Accordingly, we are persuaded that the combination of Waterson and Ross teaches the limitations of claims 2 and 12. g) Claims 3, 13, 23, and 27 Claim 3 recites “wherein the anti-key logger is installed to ensure that the anti-key logger is called first to prevent the key logging malware from operating at a more privileged access level.” Ex. 1001, 13:20–23. Claims 13, 23, and 27 depend from claims 11, 22, and 26 respectively and each claim introduces the limitation quoted above from claim 3. Ex. 1001, 14:1– 4, 14:54–57, 15:20–23. Petitioner relies on Waterson’s disclosure that it is “the first program that interfaces with the keyboard driver.” Pet. 71 (quoting Ex. 1009, 6:18–24) (emphasis removed). Petitioner argues that “[t]his means that the keylogging malware is prevented from ‘operating at a more privileged access level’ than its anti-keylogging software” and also “that it is ‘installed to ensure that the anti-keylogger is called first.’” Pet. 71 (citing Ex. 1013 ¶¶ 126–127). Patent Owner argues, as it has before, that because Waterson’s anti- key logger cannot be maintained at the most privileged access level, another program may gain priority over Waterson’s anti-key logger preventing it

50 IPR2020-00323 Patent 9,503,473 B1 from being the first program to be called upon or to interface with the keyboard driver. PO Resp. 29. Petitioner’s argument is supported by the cited evidence. Waterson discloses that its “software attempts to have exclusive control of the keyboard driver to prevent keystroke loggers from directly harvesting information.” Ex. 1009, 6:18–20. Waterson’s software accomplishes this by being “the first program that interfaces with the keyboard driver” and by ensuring “that the software filter driver is loaded before other filter drivers.” Ex. 1009, 6:20–31. Thus, we are persuaded that Waterson’s “anti-key logger is installed to ensure that the anti-key logger is called first to prevent the key logging malware from operating at a more privileged access level,” as recited in claim 3. We disagree with Patent Owner’s argument that Waterson fails to teach the limitations of claim 3 because Waterson’s software cannot be maintained at the most privileged access level. Claim 3 requires only that the anti-key logger be “installed to ensure that the anti-key logger is called first to prevent the key logging malware from operating at a more privileged access level.” As explained above, Waterson discloses that it installs its anti-key logger at the top of the driver priority list, therefore ensuring that it is “called first.” Even if Waterson’s software does not permanently remain at the top of the priority list because a key logger gains priority at a later time, it is still installed at the top of the priority list. Petitioner and Patent Owner rely on the same arguments for claims 3, 13, 23, and 27. See Pet. 74–76; PO Resp. 33. Accordingly, we are persuaded that the combination of Waterson and Ross teaches the limitations of claims 3, 13, 23, and 27.

51 IPR2020-00323 Patent 9,503,473 B1 h) Claims 4, 5, 14, and 15 Claim 4 recites “wherein the microprocessor is disposed within a computer, a mobile communication device, a smartphone, or a mobile Internet device.” Ex. 1001, 13:24–26. Claim 5 depends from claim 4 and further recites “wherein the mobile Internet device is one of a personal digital assistant (PDA), a handheld computer, a tablet computer, a laptop computer, or a notebook computer.” Ex. 1001, 13:27–30. Claims 14 and 15 depend from claim 11 and recite the same limitations as those in claims 4 and 5 respectively. Ex. 1001, 14:5–12. Petitioner relies on the following disclosure from Waterson to teach these limitations: While this anti-keystroke logging software has been developed using the Microsoft Windows operating system on a personal computer, the software employs concepts that are extendable to other devices and operating systems. Devices such as cellular phones and PDA’s all have a form of key board buffer that functions in a similar way to a common computer. Thus the buffer streaming technique is able to be implemented. Pet. 71 (citing Ex. 1009, 12:1–8; Ex. 1013 ¶¶ 128–131). Patent Owner does not separately argue claims 4, 5, 14 and 15, relying on its arguments for independent claim 1. See PO Resp. 29–30, 33. Petitioner’s argument is supported by the cited evidence. Waterson discloses that its software has been developed on a personal computer but that its software is also extendable to other devices such as phones and PDAs, as required by claims 4 and 5. Ex. 1009, 12:1–12. Accordingly, we are persuaded that the combination of Waterson and Ross teaches the limitations of claims 4, 5, 14, and 15. i) Claims 6 and 16 Claim 6 depends from claim 1 and recites “wherein the detecting and the clearing are performed in the browser.” Ex. 1001, 13:31–32. Claim 16

52 IPR2020-00323 Patent 9,503,473 B1 depends from claim 11 and recites the same limitation. Ex. 1001, 14:13–14. Petitioner argues that Waterson’s “anti-keylogging software can be ‘part of the web browser process (such as a plug-in, extension, help object, or control of the browser).’” Pet. 72 (quoting Ex. 1009, 11:14–19). Thus, Petitioner argues, that the “detecting” and “clearing” may be performed in the browser. Pet. 72 (citing Ex. 1013 ¶¶ 132–133). Patent Owner argues that in Waterson, any clearing of data occurs in the keyboard buffer as the data is in transition to the entry field of Waterson’s application. PO Resp. 31 (citing Ex. 1009, 10:49–65). According to Patent Owner, such clearing is performed at the kernel level not at the user level and that obfuscation of clean data at the kernel level does not render claim 6 obvious. PO Resp. 31 (citing Ex. 2043 ¶¶ 90–93). Patent Owner further argues that Petitioner’s positions for claim 1 and claim 6 are inconsistent, in that Petitioner relies on Ross as teaching the detection of browser form submission initiation call event for claim 1 but relies on Waterson as teaching that the detection occurs in the browser. PO Resp. 47– 48. Petitioner’s arguments are supported by the cited evidence. Waterson teaches that its anti-keylogging software can be part of the web browser process. Ex. 1009, 11:14–19. Thus, when Waterson’s software is installed as part of the web browser, it clears data in the browser. We disagree with Patent Owner’s argument that by contaminating data in the keyboard buffer Waterson does not teach clearing data in a web browser. When Waterson’s software is installed as part of the web browser, the application to which the input data is directed is the web browser’s entry field. Even when this data is contaminated by Waterson in the keyboard buffer, it is still contaminated when it arrives at the entry field of the browser

53 IPR2020-00323 Patent 9,503,473 B1 and thus is cleared in the browser. Ex. 1009, 10:56 (“The entry field 507 receives contaminated data 513.”). Moreover, in Petitioner’s combination, the clearing is performed by the anti-key logger when it is in the web browser. Thus, the clearing is performed in the browser, as required by claim 6. We also disagree that Petitioner’s positions are inconsistent between claims 1 and 6. For claim 1, Petitioner relies on Ross’s teaching of the detection of browser form submission initiation call event (Pet. 66–67 (citing Ex. 1010, 4; Ex. 1013 ¶¶ 113, 57) but incorporates this teaching into Waterson’s software, which it argues may be installed as part of the web browser (Pet. 59 (citing Ex. 1013 ¶¶ 100–108), 54–67 (citing Ex. 1010, 4; Ex. 1013 ¶¶ 112–117). Thus, the combination of Waterson and Ross teaches detecting a browser form submission initiation call event as in claim 1 and also teaches “wherein the detecting and the clearing are performed in the browser” as in claim 6. There is no inconsistency between the two. Petitioner and Patent Owner each rely on the same arguments they present for claims 6 and 16. See Pet. 74–76; PO Resp. 33. Accordingly, we are persuaded that the combination of Waterson and Ross teaches the limitations of claims 6 and 16. j) Claims 8, 20, 25, and 29 Claim 8 depends from claim 1 and recites “wherein the anti-key logger is placement aware and performs at least one check to confirm that the anti-key logger is called first.” Ex. 1001, 13:39–41. Claims 20, 25 and 29 recite the same limitation. Ex. 1001, 14:28–30, 14:62–64, 15:28–30. Petitioner argues that Waterson teaches this limitation by disclosing that it ensures that its software is at the top of the list of keyboard filter drivers by periodically verifying that this is the case. Pet. 72 (citing Ex. 1009, 6:32–

54 IPR2020-00323 Patent 9,503,473 B1 38). If not, Waterson changes the order of the list to ensure that the software filter driver is loaded before other filter drivers. Pet. 72 (citing Ex. 1009, 6:18–29, 6:32–38). Because it periodically verifies that its software is at the top of the list of keyboard filter drivers, Petitioner argues that Waterson’s software is “placement aware and performs [at least one] check to confirm that the anti-key logger is called first.” Pet. 72. Patent Owner argues that Waterson’s software does not always stay on top of the driver priority list and when it is not at the top of the list, it is unable to periodically verify that it is at the top of the list. PO Resp. 31–32 (citing Ex. 2043 ¶¶ 95–97). We are persuaded by Petitioner’s arguments, which we determine are supported by the cited evidence. Waterson discloses that its “software attempts to have exclusive control of the keyboard driver to prevent keystroke loggers from directly harvesting information” and that it is “the first program that interfaces with the keyboard driver.” Ex. 1009, 6:18–22. Waterson’s software accomplishes this by checking the order of the list of keyboard filter drivers and changing the order if it is not at the top of the list. Ex. 1009, 6:22–29. Waterson’s software periodically verifies that it remains at the top of the driver priority list. Ex. 1009, 6:32–34. This demonstrates that Waterson’s software is placement aware and checks to confirm that the anti-key logger is called first. In addition, as Petitioner points out, Patent Owner’s argument that Waterson’s software would be unable to verify that it is at the top of the list if another piece of software were to supersede it, even if true, does not address the fact that the claim limitations are met once the software checks initially to see if it is at the top of the list. Pet. Reply 14 (citing Ex. 1013 ¶¶ 134–135).

55 IPR2020-00323 Patent 9,503,473 B1 Moreover, we do not agree with Patent Owner that Waterson’s software would be unable to periodically verify its placement in the list if another program were to gain priority over it. Waterson contemplates precisely this scenario when it requests the user to reboot the system. Ex. 1009, 6:34–38 (“If at any time another program gains priority over the keyboard driver filter the software alerts the user via a pop-up window and prompts them to reboot the system.”) The evidence shows that Waterson’s software remains “placement aware” and able to “confirm that the anti-key logger is called first” even if it loses priority. Petitioner and Patent Owner rely on the same arguments for claims 8, 20, 25, and 29. See Pet. 74–76; PO Resp. 33. Accordingly, we are persuaded that the combination of Waterson and Ross teaches the limitations of claims 8, 20, 25, and 29. k) Claims 9, 21, 24, and 28 Claim 9 depends from claim 1 and recites “protecting, by the anti-key logger, the data inputs received at a physical driver level passing through a virtual keyboard level and an operating system level.” Ex. 1001, 13:42–45. Claims 21, 24 and 28 depend from claims 11, 22, and 26 respectively, and each recites limitation above from claim 9. Ex. 1001, 14:31–34, 14:58–61, 15:24–27. Petitioner argues that Waterson teaches protection at the (1) physical driver level, the (2) virtual keyboard level and (3) an operating system level, by disclosing that it prevents key stroke loggers from harvesting keystrokes (1) from the keyboard driver, (2) from the keyboard buffer and (3) from the data entry window, respectively. Pet. 73 (citing Ex. 1009, 4:22–26, 4:16–21, 4:27–30). Patent Owner argues that “[t]here is no reason in general to equate ‘keyboard buffer’ with ‘virtual keyboard level,’ and neither Petitioner nor its

56 IPR2020-00323 Patent 9,503,473 B1 expert discuss how a ‘keyboard buffer’ could be equivalent to the ‘virtual keyboard level.’” PO Resp. 32–33 (citing Ex. 2043 ¶¶ 98–103). We are persuaded by Petitioner’s arguments, which we determine are supported by the cited evidence. Dr. McDaniel provides credible testimony that a “keyboard buffer is equivalent to a virtual keyboard level.” Ex. 1024 ¶ 50. Dr. McDaniel further testifies that “[w]hen keys are pressed in any computer system, the keypresses travel from the physical keyboard driver level, through a virtual keyboard level, and onto the operating system level before reaching the application.” Ex. 1024 ¶ 51. Dr. McDaniel testifies that, because it is undisputed that Waterson’s protection begins at the physical keyboard driver level and continues up to the application level, Waterson necessarily protects data at the three claimed levels, including the virtual keyboard level. Ex. 1024 ¶ 52. Patent Owner supports its argument with the testimony of Dr. Sorini who testifies that “[i]n general, a keyboard buffer is not the same thing as a virtual keyboard level.” Ex. 2043 ¶ 99. Dr. Sorini testifies that “a virtual keyboard can be a software-based graphical window with buttons that mimics the appearance of a physical keyboard.” Ex. 2043 ¶ 100. Here, we give more weight to Dr. McDaniels’s testimony for two reasons. First, Dr. McDaniel testifies that the flow of data in a typical computer system includes keypresses going from the physical keyboard driver to the virtual keyboard level onto the operating system level and Dr. Sorini does not dispute this. Ex. 1024 ¶ 51. Second, there is no indication in the ʼ473 Patent that a “virtual keyboard level” refers to a “software-based graphical window with buttons.” Instead, the disclosure of the ʼ473 Patent appears to be more consistent with Dr. McDaniel’s interpretation rather than Dr. Sorini’s. In other words, the virtual keyboard level refers to a level more

57 IPR2020-00323 Patent 9,503,473 B1 equivalent to a keyboard buffer rather than a graphical user interface mimicking the appearance of a keyboard. Petitioner and Patent Owner rely on the same arguments for claims 9, 21, 24 and 28. See Pet. 74–76; PO Resp. 33. Accordingly, we are persuaded that the combination of Waterson and Ross teaches the limitations of claims 9, 21, 24 and 28. l) Claims 10 and 19 Claim 10 depends from claim 1 and recites “wherein the data inputs are keyed in by the user.” Ex. 1001, 13:46–47. Claim 19 depends from claim 11 and recites the same limitation. Ex. 1001, 14:26–27. For both claims, Petitioner argues that “Waterson teaches that a user ‘enter[s] data via a keyboard 501.’” Pet. 73 (citing Ex. 1009, 10:38–40, Fig. 5; Ex. 1013 ¶¶ 138–139) (alteration in original). Patent Owner does not present separate arguments for patentability of claims 10 and 19, instead relying on its arguments for claims 1 and 11 respectively. See PO Resp. 32–33. We are persuaded by Petitioner’s arguments, which we determine are supported by the cited evidence. In the same embodiment relied upon by Petitioner as teaching the limitations of claim 1, Waterson discloses that “[t]he user will enter the data via keyboard 501” which then enters the keyboard buffer. Ex. 1009, 10:38–40. Thus, the data inputs of claim 1, as taught by Waterson, are “keyed in by the user” as required by claim 10 and claim 19. m) Claim 17 Claim 17 depends from claim 11 and recites “inserting software processes for performing the detecting and clearing at the most privileged access level in the API stack of the browser.” Ex. 1001, 14:16–19.

58 IPR2020-00323 Patent 9,503,473 B1 Petitioner argues that Waterson’s software is kept at the top of the list of keyboard filter drivers which is at the most privileged access level in the API stack of the browser. Thus, according to Petitioner, Waterson teaches that the detecting and clearing processes, are performed at the most privileged access level. Pet. 77 (citing Ex. 1009, 6:18–31; Ex. 1013 ¶¶ 145–146). Patent Owner relies on the same arguments it makes for claim 6, namely that Petitioner’s positions are inconsistent between claims 11 and 17 (similar to the argument that Petitioner’s positions are inconsistent for claims 1 and 6) and that the “clearing” in Waterson is limited to data traveling from the keyboard buffer to the entry field rather than being cleared in the API stack of the browser. PO Resp. 47–49. We are persuaded by Petitioner’s arguments, which we determine are supported by the cited evidence. Waterson discloses that its “software attempts to have exclusive control of the keyboard driver to prevent keystroke loggers from directly harvesting information.” Ex. 1009, 6:18–20. Waterson’s software accomplishes this by being “the first program that interfaces with the keyboard driver” and by ensuring “that the software filter driver is loaded before other filter drivers.” Ex. 1009, 6:20–31. Waterson further teaches that its anti-key logging software can be part of the web browser process. Ex. 1009, 11:14–19. Thus, we are persuaded that Waterson’s “inserting software processes for performing the detecting and clearing at the most privileged access level in the API stack of the browser,” as recited in claim 17. We disagree with Patent Owner’s arguments for the same reasons explained above in our analysis of claim 6. See Part II.D.2.i). Accordingly, we are persuaded that the combination of Waterson and Ross teaches the limitations of claim 17.

59 IPR2020-00323 Patent 9,503,473 B1 n) Independent Claim 22 Petitioner argues that the limitations of claim 22 are similar to those of claim 1 except that claim 22 adds a limitation similar to that of claim 2 and one additional limitation. For the limitations similar to those of claims 1 and 2, Petitioner relies on the same evidence and reasoning. Pet. 74–76. The additional limitation recites “identifying form input fields on a web page having confidential data.” Petitioner argues that Waterson teaches this limitation by disclosing that “the program detects and then protects passwords on password-fields, and other data on all other entry fields.” Pet. 76 (citing Ex. 1009, 4:53–55). Petitioner argues that passwords are confidential data. Pet. 76 (citing Ex. 1013 ¶ 142). Patent Owner relies on the same arguments it made for claim 1 and claim 2. PO Resp. 13–14. As for the additional limitation of claim 22, Patent Owner argues that Waterson captures each key stroke entered by the user, and thereby captures all data and does not distinguish data as confidential based on an input field. PO Resp. 14. For the reasons explained above in our analysis of claims 1 and 2, we are persuaded by Petitioner’s arguments that the combination of Waterson and Ross teaches the limitations that are common to the three claims. We are also persuaded by Petitioner’s arguments for the additional limitation, which we determine are supported by the cited evidence. Waterson teaches that its program “detects and then protects passwords on password-fields, and other data on all other entry fields.” Ex. 1009, 4:52–55; see also id. at 11:40–41 (“The installed software is also activated when a password field is detected.”). Dr. McDaniel credibly testifies that passwords are confidential data. Ex. 1013 ¶ 142. Because Waterson teaches that it detects password

60 IPR2020-00323 Patent 9,503,473 B1 fields, we agree with Petitioner that it “identif[ies] form input fields on a web page having confidential data.” o) Independent Claim 26 Petitioner states that “claim 26 contains limitations similar to claim 22” and relies on the same arguments for both claims (which in turn rely partially on arguments made for independent claim 1 and dependent claim 2). Pet. 74–75. Petitioner provides a chart that aids in illustrating the mapping of limitations between claims, which we have reproduced below with added highlighting for ease of reference.

Pet. 75–76. In the chart, Petitioner identifies the limitations of claim 26 as corresponding to those of claim 1 and claim 2, but identifies limitation “26d” with claim 7. Limitation “26d” recites “identifying form input fields on the web page having confidential data.”

61 IPR2020-00323 Patent 9,503,473 B1 Patent Owner argues that although the Petitioner points to claim 7 as corresponding to limitation 26d, claim 7 is not challenged or addressed in the Petition under Ground 2, and thus Petitioner has failed to show with particularity why claim 26 is unpatentable. PO Resp. 49–50. Petitioner responds that it “inadvertently included a typo, mapping claim 26d (the ‘identifying’ limitation) to claim 7.” Pet. Reply 11. Patent Owner replies that allowing Petitioner to correct the error “would substantially prejudice PO, as it has filed two briefs accompanied with two declarations from its expert.” PO Sur-reply 16. Regardless, Patent Owner argues that limitation 26(d) is not rendered obvious by Waterson for the same reasons argued for the corresponding “identifying” limitation in claim 22. PO Sur-reply 16–17 (arguing that Waterson never identifies confidential data and does not differentiate confidential data from non-confidential data). Having reviewed the record, we hold Petitioner’s mistake as harmless error. The disputed “identifying” limitation of claim 26 is identical to the same limitation in claim 22. Petitioner makes clear its intention to apply the arguments made for claim 22 to claim 26 when it states that “[i]ndependent claim 11 contains limitations similar to claim 1, and claim 26 contains limitations similar to claim 22” just before presenting the chart showing the claim mappings. Pet. 74–75. Because claim 7 is not even challenged, and because Petitioner makes clear that claim 22 and claim 26 are similar, it is readily apparent that Petitioner’s chart contains an error when referencing claim 7.8

8 Petitioner’s error likely stems from copying a similar table from Ground 1, which included claim 7, to Ground 2, without fully modifying the table to account for the different claims challenged under the two grounds.

62 IPR2020-00323 Patent 9,503,473 B1 We avoid any possible prejudice to Patent Owner by applying its arguments for claim 22 to claim 26, since the two contain identical “identifying” limitations. Indeed, in its Sur-reply, Patent Owner does exactly that—applying the same arguments it made for the “identifying” limitation of claim 22 to the same “identifying” limitation of claim 26. PO Sur-reply 16–17. Having applied Patent Owner’s arguments from claim 22 to claim 26, we determine, for the reasons explained above in our analysis of claim 22, that Waterson teaches “identifying form input fields on the web page having confidential data” by disclosing that it “detects and then protects passwords on password-fields, and other data on all other entry fields.” Ex. 1009, 4:52–55; see also id. at 11:40–41 (“The installed software is also activated when a password field is detected.”). Based on its arguments for claim 1 and claim 2, in addition to the arguments for claim 22, we are persuaded by Petitioner that the combination of Waterson and Ross teaches the claim limitations of claim 26. 3. Conclusion For the reasons explained above and based on the arguments and evidence presented in the Petition, we determine that Petitioner has shown by a preponderance of the evidence that the combination of Waterson and Ross teaches each limitation of claims 1–6, 8–17, 19–29 of the ’473 Patent. Based on all of the evidence of obviousness and nonobviousness, we conclude that Petitioner has shown the unpatentability of claims 1–6, 8–17, and 19–29 as obvious over Waterson and Ross by a preponderance of the evidence.

63 IPR2020-00323 Patent 9,503,473 B1 E. Obviousness over Waterson, Ross, and Geon 1. Overview of Geon Geon relates to a computer security apparatus and a method using a security input device driver. Ex. 1012, Abstract. Data inputted through a data input unit is directly encrypted at the security input device driver without passing through an operating system (“OS”) input device driver supporting in an OS. Ex. 1012, Abstract. More specifically, a security input interrupt service routine (“ISR”) replaces a conventional input ISR, where the security input ISR protects keyboard data. Ex. 1012, 1:15–19, 5:45–50, 6:11–13. In the security input ISR, keyboard input data is read from a keyboard buffer through a 60h port, and stored. Ex. 1012, 7:1–20. The keyboard buffer is then reset to dummy data by calling the “Clearbuffer” function to prevent the initial input data from being read through the 60h port. Ex. 1012, 7:1–17. Further, the keyboard input data is directly encrypted at a security input device driver without passing through an OS input device driver. Ex. 1012, 7:61–64. 2. Combination of Waterson, Ross, and Geon Petitioner argues that to the extent the Board finds that Waterson and Ross do not disclose the “clearing” limitation (limitation 1.d) under the construction agreed upon by the parties in the underlying district court litigation, Geon “provides an express disclosure of ‘clearing’ confidential data under any definition of that term.” Pet. 77. Petitioner, therefore, combines Geon’s function of clearing user inputted data out of a keyboard buffer with the combined key logging software of Waterson and Ross in order to provide another level of protection against keylogging malware. Pet. 79 (citing Ex. 1013 ¶ 148).

64 IPR2020-00323 Patent 9,503,473 B1 Petitioner argues the motivation to combine Geon’s clearing function with the combined keylogging software of Waterson and Ross is to provide additional protection against keylogging malware, for reasons similar to the reasons discussed with respect to Waterson and Ross. Pet. 79 (citing Ex. 1013 ¶¶ 102, 148). As argued by Petitioner, Geon, like both Waterson and Ross, teaches anti-key logging software that provides protection against key logging malware at the keyboard driver level. Pet. 79. Petitioner further argues that, similar to the ’473 Patent, Waterson, and Ross, Geon is directed to preventing capture of sensitive information by malicious key logging software. Pet. 80. As further argued by Petitioner, a person of ordinary skill in the art would have had a reasonable expectation of success in incorporating the clearing function described in Geon into the Waterson/Ross system because Geon describes a method similar to that described in Waterson. Pet. 81. For example, in Geon, the keyboard input is stored while the Clearbuffer function sends dummy data to the 60h port. Pet. 81 (citing Ex. 1012, 7:1– 25). This is similar to Waterson, which, according to Petitioner, stores clean data while streaming random data to the keyboard buffer. Pet. 81 (citing Ex. 1009, 10:38–57; Ex. 1013 ¶ 152). Petitioner argues that incorporating Geon’s functionality into the Waterson/Ross system would have been a routine modification because both systems send random or dummy data to the keyboard buffer so that the keystroke loggers receive contaminated data. Pet. 81–82 (citing Ex. 1009, 10:49–57; Ex. 1012, 7:4–15). According to Petitioner, one of ordinary skill in the art would have understood that Geon’s Clearbuffer function would have been desirable to incorporate into Waterson to empty out the keyboard buffer after data input is complete, thereby increasing security. Pet. 82 (citing Ex. 1012, 7:4–14; Ex. 1013 ¶ 154).

65 IPR2020-00323 Patent 9,503,473 B1 Patent Owner argues that the proposed combination fails to explain how one of ordinary skill in the art would go about combining the disparate elements of Geon into the Waterson and Ross system, including how Geon’s Clearbuffer function would function with Ross’s teaching of trapping the form submission event. PO Resp. 54 (citing Ex. 1043 ¶¶ 122–126). Patent Owner argues that Petitioner has failed to establish that a skilled artisan would have had a reasonable expectation of success in combining Geon with Waterson and Ross and that Petitioner has not presented the obviousness combination with the requisite particularity. PO Resp. 55–56. Patent Owner further argues that incorporating Geon’s Clearbuffer function would render Waterson inoperable as the entry field of Waterson would no longer be able to separate clean data from random data because the entry field would no longer receive contaminated data and the underlying application would no longer receive clean data. PO Resp. 54 (citing Ex. 1012, 6:55–65, 7:1–15, 10:48–65; Ex. 2043 ¶¶ 127–133, 138–142). Patent Owner argues that with the incorporation of Geon, the keyboard buffer will be reset before the clean and random data is transmitted to the entry field. PO Resp. 54–55. Patent Owner argues that even with Geon’s incorporation, the clearing would occur at the keyboard buffer, but the claimed invention is directed to clearing in the browser. PO Resp. 55 (citing Ex. 1012, 7:1–15, 10:48–55). We have carefully considered all arguments and supporting evidence regarding the rationale for combining the teachings of Geon with Waterson and Ross. We are persuaded by Petitioner that one of ordinary skill in the art would have been motivated to combine the cited prior art to arrive at the inventions recited in the challenged claims. Dr. McDaniel provides credible testimony that a person of ordinary skill would have been motivated to

66 IPR2020-00323 Patent 9,503,473 B1 combine Geon’s Clearbuffer function with the anti-key logging software of Waterson and Ross to provide another level of protection and to address the principle of “complete mediation.” Ex. 1013 ¶ 148. Dr. McDaniel also credibly testifies that the addition of the Clearbuffer function would help clear memory of sensitive data to prevent some other program under control of an adversary to read the sensitive data after it has been used. Ex. 1013 ¶ 150. Given that both Geon and Waterson are directed to preventing capture of sensitive information by malicious key logging software, we agree with Petitioner that one of ordinary skill in the art would have been motivated to use Geon’s Clearbuffer function in the Waterson/Ross combination to provide additional keystroke protection. We disagree with Patent Owner that Petitioner has not established a reasonable expectation of success in combining Geon with Waterson and Ross. Dr. McDaniel credibly testifies that a skilled artisan would have had a reasonable expectation of success in incorporating Geon’s Clearbuffer function into the Waterson/Ross system because the method in Geon is similar to that in Waterson. Ex. 1013 ¶ 152. Specifically, Geon teaches storing the keyboard input data and then sending dummy data to the 60h port, and Waterson teaches storing the clean data while streaming random data to the keyboard buffer. Ex. 1013 ¶ 152 (citing Ex. 1012, 7:1–25; Ex. 1009, 10:38–57). Because Geon’s Clearbuffer function is similar to Waterson’s own method of using random data to contaminate keyboard inputs, we find that one of ordinary skill in the art would have had a reasonable expectation of success in using Geon’s method with Waterson and Ross. We also disagree that Petitioner has not presented the obviousness combination with the requisite particularity. Patent Owner’s argument is

67 IPR2020-00323 Patent 9,503,473 B1 based on Petitioner’s apparent inconsistency in relying on the combination of Geon, Waterson, and Ross in Ground 3 while stating that claims 11–16 and 19–26 are rendered obvious by Waterson and Ross for the reasons set forth in Ground 2. PO Resp. 56–57; Pet. 85. It is clear, however, that Petitioner’s intent here is to take the same approach it did in Ground 2, by relying on its contentions for claim 1 and applying them to independent claims 11, 22, and 26, rather than repeating those contentions again.9 See Pet. 74–76. Thus, on the whole, it is clear that Petitioner relies on the combination of Geon, Waterson and Ross for claims 11–16 and 19–26 with Geon being relied on for the “clearing” limitation. We disagree that Petitioner that incorporating Geon’s Clearbuffer function would render the Waterson/Ross combination inoperable for its intended purpose. Patent Owner argues that using the Clearbuffer function of Geon would reset the keyboard data before it is sent to Waterson’s entry field and therefore the entry field would no longer be able to receive the clean data. PO Resp. 54. But this argument ignores that in Petitioner’s combination the clearing of buffers occurs after data input is complete. See Pet. 82; Pet. Reply 23. Thus, the entry field and underlying application would receive data as before and the buffers would be cleared once the data was received by each. Finally, Patent Owner’s argument that in Petitioner’s combination the clearing would occur at the keyboard buffer, while the claimed invention is

9 This intention is even clearer when viewed in conjunction with Dr. McDaniel’s declaration, which states that “claim 11 contains limitations similar to claim 1, and claim 26 contains limitations similar to claim 22. Claim 22’s limitations are a combination of limitations similar to those in claim 1 and its dependent claims 2 and 7.” Ex. 1013 ¶ 169.

68 IPR2020-00323 Patent 9,503,473 B1 directed to clearing in the browser has been addressed in our analysis of claim 1 and 2 under Ground 2 above. See supra Part II.D.2.e)(2); Part II.D.2.f). 3. Analysis of the Limitations of Claim 1 For its arguments regarding independent claim 1 as obvious over Waterson, Ross, and Geon, Petitioner relies largely on the same arguments set forth above regarding claim 1 as obvious over Waterson and Ross. See Pet. 83, 85. Petitioner cites to Geon, however, for teaching “clearing, by the anti-key logger, confidential data from the data inputs to protect against the threat of key logging malware capturing the confidential data,” as recited in claim 1. Pet. 83–84. As explained above, Petitioner relies on Geon for an express teaching of the “clearing” limitation that it argues would teach the limitation under any definition of that term, not just the one agreed upon by the parties in the underlying litigation. Pet. 77. Patent Owner does not present arguments specifically directed to Petitioner’s contentions regarding the limitations of claim 1 under Ground 3, relying rather on the same arguments it made with respect to Ground 2. We are persuaded by Petitioner’s argument that Geon’s Clearbuffer function combined with Waterson and Ross teaches the claimed “clearing, by the anti-key logger, confidential data from the data inputs to protect against the threat of key logging malware capturing the confidential data.” Dr. McDaniel provides credible testimony that Geon teaches a function that resets that 60h port to “dummy data” by use of a Clearbuffer function that clears the output buffer. Ex. 1013 ¶¶ 159–160 (citing Ex. 1012, 7:1–17). Dr. McDaniel testifies that a person of ordinary skill in the art would have recognized that Geon’s Clearbuffer function could be utilized to clear confidential data from the keyboard buffer disclosed in Waterson. Ex. 1013

69 IPR2020-00323 Patent 9,503,473 B1 ¶ 160. We agree with Dr. McDaniel that under the agreed upon construction of the “clearing” limitation, Geon’s Clearbuffer function teaches “clearing, by the anti-key logger, confidential data from the data inputs to protect against the threat of key logging malware capturing the confidential data,” as recited in claim 1 because it would remove the meaning from the inputted data. For the reasons explained above in our analysis of claim 1 under Petitioner’s Ground 2 allegations over Waterson and Ross, we also agree with Petitioner that the combination of Geon, Waterson, and Ross teaches the remaining limitations of claim 1. 4. Claims 2–6, 8–17, and 19–29 Petitioner contends claims 2–6, 8–17, and 19–29 of the ʼ473 Patent are unpatentable under 35 U.S.C. § 103 as obvious over Waterson, Ross, and Geon and provides specific arguments for each challenged claim, relying on the same evidence and reasoning as those relied upon in the previous grounds while relying on Geon a teaching the “clearing” limitation as discussed above. See Pet. 84–85. Patent Owner relies on the same arguments as discussed previously in Grounds 2 and 3. We determine Petitioner presents sufficient evidence to establish that the combination of Waterson, Ross, and Geon teaches the limitations of claims 2–6, 8–17, and 19–29. 5. Conclusion For the reasons explained above and based on the arguments and evidence presented in the Petition, we determine that Petitioner has shown by a preponderance of the evidence that the combination of Waterson, Ross and Geon would have taught each limitation of claims 1–6, 8–17, and 19–29 of the ’473 Patent. Based on all of the evidence of obviousness and

70 IPR2020-00323 Patent 9,503,473 B1 nonobviousness, we conclude that Petitioner has shown the unpatentability of claims 1–6, 8–17, and 19–29 as obvious over Waterson, Ross, and Geon by a preponderance of the evidence.

III. CONCLUSION10 Weighing all of the evidence of obviousness and nonobviousness together, including the content of the prior art, the differences between the prior art teachings and the claim limitations, and the objective indicia of nonobviousness, we determine that Petitioner has shown by a preponderance of the evidence that claims 1–6, 8–17, and 19–29 would have been obvious over the combination of Waterson and Ross, and the combination of Waterson, Ross, and Geon. We determine that Petitioner has not shown by a preponderance of the evidence that claims 1–29 are anticipated by the ʼ445 Patent. In summary: Claims Claims 35 Reference(s)/ Shown Not Shown Claims U.S.C. § Basis Unpatentable Unpatentable 1–29 102(b) the ʼ445 Patent 1–29 1–6, 8–17, 103(a) Waterson, 1–6, 8–17, 19–29 Ross 19–29

10 Should Patent Owner wish to pursue amendment of the challenged claims in a reissue or reexamination proceeding subsequent to the issuance of this decision, we draw Patent Owner’s attention to the April 2019 Notice Regarding Options for Amendments by Patent Owner Through Reissue or Reexamination During a Pending AIA Trial Proceeding. See 84 Fed. Reg. 16,654 (Apr. 22, 2019). If Patent Owner chooses to file a reissue application or a request for reexamination of the challenged patent, we remind Patent Owner of its continuing obligation to notify the Board of any such related matters in updated mandatory notices. See 37 C.F.R. §§ 42.8(a)(3), (b)(2).

71 IPR2020-00323 Patent 9,503,473 B1 1–6, 8–17, 103(a) Waterson, 1–6, 8–17, 19–29 Ross, Geon 19–29 Overall 1–6, 8–17, 7, 18 Outcome 19–29

IV. ORDER In consideration of the foregoing, it is hereby: ORDERED claims 1–6, 8–17, 19–29 of the ʼ473 Patent are held to unpatentable; and FURTHER ORDERED that, because this is a Final Written Decision, the parties to the proceeding seeking judicial review of the decision must comply with the notice and service requirements of 37 C.F.R. § 90.2.

72 IPR2020-00323 Patent 9,503,473 B1 FOR PETITIONER:

James M. Glass John T. McKee QUINN EMANUEL URQUHART & SULLIVAN, LLP [email protected] [email protected]

FOR PATENT OWNER:

Amardeep S. Grewal Sidharth Kapoor REED SMITH LLP [email protected] [email protected]

73