5 Tribunals' Rules To Help Patent Litigators Avoid AI Disasters

Artificial intelligence will present information with confidence, even if it is hallucinating.[1] Use it as an aid, but always verify its outputs.

Patent litigators — given their technical backgrounds — are uniquely poised to leverage AI tools in their legal practices. However, they must be cautious about overreliance on AI to avoid their, or their expert witness's, hard-earned credibility being shattered by an AI-fabricated citation.

In a recent case in the U.S. District Court for the District of Minnesota, where an expert's report included one such hallucinated citation, U.S. District Judge Laura M. Provinzino pulled no punches: In January, she excluded the expert's report entirely and scolded the attorney regarding his nondelegable duty under Rule 11 of the Federal Rules of Civil Procedure to independently verify his submissions.[2]

Given the critical role of expert witnesses in a patent case, patent litigators would be wise to scrutinize their experts' reports as much as they do their own court filings.

Courts are expecting patent attorneys to keep abreast of AI technology and at minimum, have an understanding of AI's propensity to hallucinate. The expectations may be even higher for patent litigators, who are presumably more tech-savvy than lawyers in other practice areas. To aid in these efforts, we provide the following survey of AI rules in several patent-heavy fora.

Survey of Five Tribunals' Standing Orders on AI

Courts' and agencies' AI-related rules typically focus on attorneys' duties under Rule 11.[3] However, some tribunals provide more detailed AI requirements through local rules or judicial standing orders. For example, some tribunals are requiring disclosure of filings that were "drafted" by generative AI. While AI guidance comes from multiple sources, including bar associations, clients and law firms, litigants may be wise to focus on rules issued by tribunals and judges to avoid the risk of sanctions.

The following sections highlight AI-related rules from patent-heavy tribunals, with which patent litigators will want to be most familiar. Neither the District of Delaware nor its judges currently have AI-related rules.

Northern District of California

Three judges' AI-related standing orders in the U.S. District Court for the Northern District of California provide a sampling of the range of approaches taken by courts and judges. U.S. District Judge Rita F. Lin's standing orders remind counsel that they "alone bear[] ethical responsibility for all statements made in filings," including "the accuracy of any Paul Browning Elizabeth Ferrill Abhinav Garg research conducted by" AI.[4] This duty extends to the numerous and lengthy expert reports filed in each patent case.

U.S. District Judge Araceli Martínez-Olguin steps it up significantly, requiring that "specifically designated lead trial counsel" must certify that they have "personally verified" the accuracy of AI-generated content.[5] Omitting the certification statement "will be grounds for sanctions."[6] Counsel must also save their AI prompts "in the event those records become relevant at any point."[7] Of note, the public sanctions order issued in the 2023 Mata v. Avianca case in the U.S. District Court for the Southern District of New York included screenshots of the attorney's AI prompts.[8]

U.S. Magistrate Judge Peter H. Kang addresses AI from the start of a case, requiring parties to meet and confer about AI's role in the case and raise any issues during the joint case management conference.[9] Judge Kang takes a nuanced approach to formulating rules related to AI tools.

For example, his rules exempt AI tools that only tangentially relate to the practice of law, such as those used for accounting or business development and similar back-office tools, as well as "traditional legal research, word processing, spellchecking, grammar checking, or formatting software tools."[10] Thus, attorneys do not need to worry where AI is not generating the substance of a filing. This nuanced guidance exemplifies practical rulemaking: Blanket rules on "AI tools" will unjustifiably cover many legal research and word processing tools, wherein AI is deeply embedded.

Regarding generative AI tools, Judge Kang requires counsel to have "competent training, knowledge, and understanding of the limitations and risks" of using AI, including staying current on relevant case law and legal standards.[11] Counsel also must disclose when a paper was "created or drafted with any use of an AI tool" and confirm the accuracy of AI-generated legal citations, or else risk potential sanctions.[12]

Counsel must also identify evidence that they have reason to believe was at least partially created by AI and serve notice to opposing counsel.[13] On confidentiality, the judge cautions that using third-party AI tools could risk confidential information covered by a protective order.[14] Thus, he requires saving all "all prompts or inquiries submitted to any such third-party AI tools."[15]

Stringent AI reporting requirements, while not an outright ban, may have a chilling effect on its use.[16] Some commentators have noted that the array of AI-related rules issued by tribunals and judges may dull the democratization of legal access that might otherwise be unlocked by AI tools.[17]

Of the rules illustrated in this section, Judge Martínez-Olguin's set of rules are the most likely to dissuade attorneys from using AI tools. For example, faced with the prospect of personally verifying all legal and factual assertions, lead trial counsel may instead choose not to use any AI tools. Counsel may also be averse to saving and potentially disclosing their team's AI prompts, which may include privileged client information, and thus forgo the use of any AI tools for this reason as well.

Of the variety of approaches taken by the Northern District of California's individual judges, some combination will prove to minimize the burden on litigants while adequately safeguarding the court's time and attention. Until then, litigants will need to pay close attention to their assigned judge's standing orders.

District of New Jersey

In the U.S. District Court for the District of New Jersey, U.S. District Judge Evelyn Padin's disclosure requirements are particularly relevant for patent litigators, as accuracy and reliability of technical content are critical in intellectual property cases. She requires that counsel disclose the AI program used and the portion of the filing drafted by AI and certify that the AI-generated material was "diligently reviewed by a human being for accuracy and applicability."[18]

Regarding disclosure of the AI program, some courts may view the use of general-purpose AI tools, such as ChatGPT, with a more jaundiced eye than certain legal-specific AI tools, such as Lexis+ Brief Analysis or Harvey AI.[19]

Eastern District of Texas

The U.S. District Court for the Eastern District of Texas' local rule cautions lawyers to not replace their "most important asset — the exercise of independent legal judgment" with that of an AI tool.[20] Counsel may use AI tools in their work product if they (1) believe, in their professional legal judgment, that the client is best served by it; and (2) verify all AIgenerated content.[21]

No judges on the court have issued AI-related standing orders.

Western District of Texas

While the U.S. District Court for the Western District of Texas does not have AI-related local rules, U.S. District Judge Fred Biery has issued a short statement on the use of AI, reminding attorneys of their obligations under Rule 11.[22]

USPTO

The U.S. Patent and Trademark Office has published two rounds of AI guidance directed to litigants practicing before the agency. The February 2024 memo reminds practitioners that "[s]imply assuming the accuracy of an AI tool is not a reasonable inquiry" under Rule 11.[23] Further, presenting hallucinated cases or facts may qualify as presenting a paper for an "improper purpose because it could 'cause unnecessary delay or needless increase in the cost'" of the proceeding.[24]

Sanctions include striking the offending paper, precluding the practitioner or party from submitting papers, terminating the proceeding, disciplinary action, and even criminal liability for "knowing and willful violations."[25]

The USPTO's April 2024 Federal Register publication adds that practitioners may use AI tools if they review all generated content,[26] personally sign all papers, sponsor only human support staff in Patent Center, avoid data mining within Patent Center, and pay close attention to the AI tool's confidentiality policies — essential steps for patent litigators handling sensitive intellectual property matters.[27]

Practical Tips for Patent Litigators

Patent attorneys seeking to utilize AI drafting tools should consider starting small by generating claim charts before moving to easily verifiable content, such as procedural histories. With its creative lingual abilities, AI may also be helpful when brainstorming claim interpretations. Because AI-related rules vary, it is wise to always check for the most recent versions of the relevant forum's rules and to consult local counsel for unspoken expectations.[28]

Patent-specific verification is crucial, especially for technical assertions or prior art citations. Attorneys should consider developing systematic protocols to validate AI-generated technical claims and legal citations against reliable patent databases, and cross-check assertions with actual exhibits or expert analysis.

Using a secure, vetted AI platform specifically designed for patent litigation or technical document review can also significantly mitigate confidentiality and accuracy risks. Documenting each verification step safeguards against potential mistakes and preserves client confidence.

Ultimately, patent litigators can reap significant time savings by selectively applying AI, provided they pair it with rigorous human oversight. When these tools are harnessed thoughtfully, they may enable attorneys to focus on strategy, in-depth legal research and persuasive argumentation — the hallmarks of good advocacy.

Endnotes

[1] AI Hallucinations refer to plausible-sounding but false information generated by AI. Because AI is broadly trained on the internet, it is prone to asserting false information as fact.

[2] Order Granting In Part and Denying In Part Plaintiffs' Motion to Exclude Expert Testimony and Denying Defendant's Motion For Leave to File an Amended Expert Declaration, Kohls v. Ellison, No. 24-cv-3754 (LMP/DLM) (D. Minn. Jan. 10, 2025).

[3] See Fed. R. Civ. P. 11. Rule 11 states, in part, that by filing a paper with the court, the attorney is representing that to the best of their knowledge and after reasonable inquiry, the papers' factual contentions have evidentiary support and all "legal contentions are warranted by existing law or by a nonfrivolous argument for extending, modifying, or reversing existing law or for establishing new law," else risk sanctions.

[4] Rita F. Lin, Standing Order for Civil Cases Before Judge Rita F. Lin at 5, U.S. Dist. Ct. N.D. Cal. (Nov. 9, 2023).

[5] Araceli Martínez-Olguín, Standing Order for Civil Cases Before District Judge Araceli Martínez-Olguín at 5, U.S. Dist. Ct. N.D. Cal. (Nov. 22, 2023).

[6] Id.

[7] Id.

[8] See Opinion and Order on Sanctions, Mata v. Avianca, No. 22-cv-1461 (S.D.N.Y. June 22, 2023).

[9] See Peter H. Kang, Standing Order for Civil Cases Before Magistrate Judge Peter H. Kang at 2, U.S. Dist. Ct. N.D. Cal. (Jul. 14, 2023).

[10] Id. at 9.

[11] Id. at 9–10.

[12] Id.

[13] Id.

[14] Id. at 12.

[15] Id.

[16] Some commentators have criticized complex and demanding AI-related rules as having a chilling effect, as attorneys may be "more likely to say, 'Just don't use it.'" See Sarah Martinson, Judges And Law Scholars Divided Over AI Standing Orders, LAW360 (Mar. 22, 2024), https://www.law360.com/pulse/articles/1810715/judges-and-law-scholars-divided-over-ai-standing-orders; see also Michael Berman, Should Courts Use Standing Orders or Local Rules to Address A.I.?, JD Supra (Dec. 4, 2023), https://www.jdsupra.com/legalnews/should-courts-use-standing-orders-or-3452888/; Shweta Watwe, Judges Reflect on GenAI Use One Year After ChatGPT's Debut, Bloomberg L. (Nov. 28, 2023), https://news.bloomberglaw.com/litigation/judges-reflect-on-genai-use-one-year-after-chatgpts-debut.

[17] See John G. Roberts, Jr., 2023 Year-End Report on the Federal Judiciary at 5 (Dec. 31, 2023), www.supremecourt.gov/publicinfo/year-end/2023year-endreport.pdf.

[18] See Judge Evelyn Padin's General Pretrial And Trial Procedures at 2 (revised Nov. 13, 2023).

[19] The authors of this article express no opinion about the quality of any of these AI tools other than that ChatGPT is directed to a general audience and Lexis Brief Analysis and Harvey AI are directed to attorneys.

[20] Rodney Gilstrap, General Order Amending Local Rules at 4, U.S. Dist. Ct. E.D. Tex. (Oct. 30, 2023).

[21] Id.

[22] Court Advisory Concerning Discovery and Other Matters, Montez v. Esparza, No. SA-23-CV-1483-FB (W.D. Tex. Dec. 6, 2023).

[23] Katherine K. Vidal, The Applicability of Existing Regulations as to Party and Practitioner Misconduct Related to the Use of Artificial Intelligence at 3, U.S. Pat. Trademark Off. (Feb. 6, 2024).

[24] Id. at 4.

[25] Id.

[26] Guidance on Use of Artificial Intelligence-Based Tools in Practice Before the United States Patent and Trademark Office, 89 Fed. Reg. 25609 (Apr. 11, 2024).

[27] Id. at 12611–17.

[28] Online trackers of courts' AI-related rules can be a helpful place to start, but may not always be current. See, e.g., Tracking Federal Judge Orders On Artificial Intelligence, LAW360, https://www.law360.com/pulse/ai-tracker.