O USPTO adverte contra a confiança cega na inteligência artificial
U.S. Patent and Trademark Office (USPTO) Director Kathi Vidal recently released a memorandum on the subject of the use of artificial intelligence (AI) by parties during proceedings before the Trademark Trial and Appeal Board (TTAB) and the Patent Trial and Appeal Board (PTAB).[1] In the memo, Director Vidal warns that parties are responsible for the content of their filings before the Boards, even when assisted by AI.
Concerns About AI Misuse
Director Vidal begins by recognizing that AI has already presented challenges for judges in other forums.[2] While Chief Justice John Roberts has observed that AI “has great potential to dramatically increase access to key information for lawyers and non-lawyers alike,” AI may also present false information as fact.[3] For example, in a widely reported incident, in the Southern District of New York a lawyer turned in a brief including citations to cases that were made up or “hallucinated” by a popular AI tool.[4] Last year, two attorneys were fined for their use of hallucinated cases in a brief.[5] Director Vidal expresses concerns that misuses of AI will add delays and incur unnecessary costs on parties before the TTAB and PTAB.[6]
Responsibility Lies on the Parties
Director Vidal notes that the USPTO has rules in place to prevent parties from engaging in misconduct. The USPTO Rules of Professional Conduct require that any signatory to a submission to either of the Boards certifies inter alia “that all statements made therein of the party’s own knowledge are true, that any legal contentions are warranted by existing law or by a nonfrivolous argument for the extension…or reversal of existing law, and that factual contentions have evidentiary support.”[7] The memo further notes that it is not enough to assume any AI tools used to create a submission to the PTO are providing correct information. All signatories to a filing thus have a duty to ensure that the filing meets the criteria stated in the rules. Practitioners are also reminded that sanctions are available under the rules, from “[s]triking the offending paper” up to “[t]erminating the proceedings in the Office.”[8] Knowing and willful violations could result in criminal liability.[9]
Conclusões
The recent memo from Director Vidal makes it clear that all parties and practitioners appearing before USPTO Boards have a duty to ensure that all information presented in their filings is factually accurate and that any arguments pursued are based on valid legal positions. Practitioners and parties should be aware of the limitations of AI tools and should not assume that the material generated by them is fit for submission in a legal proceeding. Before submitting a filing to a Board, parties and practitioners should take extra care to review any material generated with the assistance of AI tools to ensure that the AI-generated material is indeed accurate. Of particular concern should be citations to legal cases, which should be thoroughly checked to ensure that the cases really exist and stand for the stated legal principle — of course, this remains good practice even when AI-generative tools are not used for a given filing.
The USPTO recognizes that the use of AI tools to generate material in the legal industry is becoming increasingly common.[10] While AI tools have the potential to be incredibly helpful, these tools must be used judiciously and with an understanding that the ultimate responsibility for a filing lies with the people appearing before the Boards.
[1] “The Applicability of Existing Regulations as to Party and Practitioner Misconduct
Related to the Use of Artificial Intelligence” (February 6, 2024).
[2] Id. at 1.
[3] Id. at 1 (quoting 2023 Year-End Report on the Federal Judiciary at 5-6, www.sup remecourt.gov/publicinfo/
year-end/2023year-endreport.pdf (Dec. 31, 2023).
[4] Pranshu Verma, “Michael Cohen Used Fake Cases Created by AI in Bid to End His Probation,”
Washington Post, www.washingtonpost.com/technology/2023/12/29/michael-cohen-ai-googlebard-
fake-citations (Dec. 29, 2023).
[5] Dan Milmo et al., “Two US lawyers fined for submitting fake court citations from ChatGPT,” The Guardian, https://www.theguardian.com/technology/2023/jun/23/two-us-lawyers-fined-submitting-fake-court-citations-chatgpt (Jun. 23, 2023).
[6] “The Applicability of Existing Regulations as to Party and Practitioner Misconduct
Related to the Use of Artificial Intelligence” (February 6, 2024) at 2.
[7] Id. at 3 (quoting 37 C.F.R. §§ 11.18(b)(1), 11.18(b)(2), 11.18(b)(2)(ii), 11.18(b)(2)(iii) (internal quotes removed).
[8] Id. at 4 (citing 37 C.F.R. §§2.119(e), 2.193, 42.6(a)(4), 42.l1, 42.12; Trademark Trial and Appeal
Board Manual of Procedure §527.03).
[9] Id. (referring to 37 C.F.R. § l1.l8(b)(1), (c)).
[10] Indeed, an initial draft of this blog post was based on a summary of the memo provided by an AI tool. All references were thoroughly reviewed for accuracy.
