Director Issues Guidance on Use of AI for Documents Submitted to PTAB | Jones Day
USPTO Director Katherine K. Vidal issued guidance on February 6, 2024 regarding the use of AI in drafting materials submitted to the USPTO’s administrative boards, including the PTAB. Of biggest concern in the guidance is the submission of false information to the PTAB possibly generated by AI. In the guidance, Director Vidal discusses certain USPTO Rules that practitioners are required to follow and identifies sanctions for violating same. Although the guidance highlights certain USPTO Rules relevant to the use of AI, Director Vidal cautions that these are just “exemplary and not exhaustive.” Regardless of whether AI is used to generate PTAB submissions, the buck stops with the practitioner signing the submitted papers.
Director Vidal acknowledges that the use of AI in matters before the PTAB “poses opportunities to expand access and lower costs,” while at the same time “poses significant concerns … that AI will be misused or left unchecked.” The guidance made reference to U.S. Supreme Court Chief Justice John Roberts’s discussion about the use of artificial intelligence in his 2023 year-end report. Relevant to judicial proceedings, Justice Roberts acknowledged that AI has great potential to increase access to the legal system, but also cautioned against its privacy risks and potential for biases. One of the shortcomings of AI identified by Justice Roberts is “hallucination.” Hallucination causes AI to present fabricated information such as non-existent cases. Justice Roberts also predicted that while AI will significantly affect judicial work, human judges will be around for the foreseeable future as legal determinations often involve gray areas, nuances, or fact-specific resolutions that require the application of human judgment.
Tying back to the challenges raised by the Chief Justice Roberts, Director Vidal’s guidance stated that when it comes to those challenges, “[i]t is expected that staff in the PTAB … will successfully apply their existing skills and relevant existing rules.” The guidance highlights that the USPTO “already has experience with and rules addressing” misconduct to include, for example, human citations to irrelevant sources. Such rules include the USPTO Rules of Professional Conduct and the signatory requirement within 37 CFR 11.18 for submissions made to the USPTO. Under Rule 11.18, the signatory certifies a number of things, including that statements made within a submitted paper are true and that after a reasonable inquiry any legal contentions are supported by existing law (or a nonfrivolous argument for extension or reversal of existing law). Director Vidal states that “[s]imply assuming the accuracy of an AI tool is not a reasonable inquiry.”
To no surprise, the guidance also highlights a number of sanctions practitioners can face should any paper submitted not comply with USPTO Rules. These sanctions range from striking the offending paper or precluding its submission to criminal liability in cases of knowing and willful violations.
Takeaway: In recent guidance, Director Vidal cautioned that the USPTO knows how to deal with erroneous citations and expects PTAB judges will continue to apply their existing skills in that regard to AI. Practitioners signing USPTO submissions must make a reasonable inquiry into the truth of any paper submitted. Assuming that an AI tool is accurate is not a reasonable inquiry under Rule 11.18. Practitioners can be subjected to a number of sanctions if a document is submitted that violates the USPTO rules, regardless if AI was used to prepare such document.
