The Fifth Circuit is the latest court to address potential misuse of generative artificial intelligence (GenAI). The court is seeking comments on an amendment to Rule 32.3 that would require litigants to certify that they reviewed and verified any GenAI-generated content in their filings. Failure to comply could result in striking the document or sanctions.
The text of the proposed rule is below (with changes in bold):
32.3. Certificate of Compliance. See Form 6 in the Appendix of Forms to the Fed. R. App. P. Additionally, counsel and unrepresented filers must further certify that no generative artificial intelligence program was used in drafting the document presented for filing, or to the extent such a program was used, all generated text, including all citations and legal analysis, has been reviewed for accuracy and approved by a human. A material misrepresentation in the certificate of compliance may result in striking the document and sanctions against the person signing the document.
The rule will also alter Form 6, the Certificate of Compliance. Now, litigants will have to check a box certifying that in preparing the relevant filing (1) they did not use GenAI or (2) they used GenAI but “all generated text, including all citations and legal analysis,” was “reviewed for accuracy and approved by a human.” The proposed language is in bold:
3. This document complies with the AI usage reporting requirement of 5th Cir. R. 32.3 because:
- no generative artificial intelligence program was used in the drafting of this document, or
- a generative artificial intelligence program was used in the drafting of this document and all generated text, including all citations and legal analysis, has been reviewed for accuracy and approved by a human.
The Fifth Circuit is the first federal appeals court to consider a GenAI-specific rule. But its proposal comes on the heels of a flurry of lower court standing orders aimed at addressing similar concerns. In May, Judge Brantley Starr of the Northern District of Texas issued a standing order requiring litigants to certify that any language drafted by GenAI was “checked for accuracy, using print reporters or traditional legal databases, by a human being.” The Eastern District of Texas followed suit, proposing a similar local rule. And at least one Texas state court now requires litigants to submit a form certifying that they have verified any AI-generated legal research or language prior to submitting it or using it in a proceeding.
These rules likely represent a response to recent concerns about litigants’ reliance on GenAI-generated legal research. Earlier this year, a viral SCOTUSblog article highlighted ChatGPT’s pitiable performance when asked fifty questions about the U.S. Supreme Court. When prompted to name three noteworthy opinions by Justice Ruth Bader Ginsburg, the program confidently — and incorrectly — asserted that the justice dissented in the landmark 2015 Obergefell v. Hodges decision. Additionally, several lawyers recently made headlines when they were sanctioned for briefing nonexistent case law that ChatGPT had “hallucinated.”
It is too soon to tell whether new court rules will prevent the abuse of GenAI seen in recent months. To be sure, threat of sanctions is certain to make lawyers think twice before blindly relying on GenAI-generated case cites. But it is also unclear that these issues necessitate formal certification. Indeed, preexisting statutes and ethical rules already require lawyers to use technology responsibly and to ensure their filings’ accuracy.
Moreover, forcing lawyers to publicly disclose use of GenAI tools could come at a cost. Certification requirements may discourage lawyers from utilizing new GenAI technologies that are specifically aimed at enhancing legal practice — and therefore present less serious validity concerns. For example, unlike ChatGPT, GenAI programs like Casetext, Fastcase and Bloomberg Brief Analyzer are specifically designed to expedite and simplify legal research. Moreover, OpenAI recently announced its rollout of customizable, internal-only “GPT” programs that companies can mold to specific purposes and feed specialized data sets. Because these programs could be tailored toward specific functions, they may be more reliable than current, more mainstream GenAI technologies. But even if they are highly effective, lawyers may be reluctant to take advantage of them if they are required to disclose their use in court filings.
Additionally, as these technologies proliferate, it may be difficult for lawyers to even discern whether they have used GenAI technology. For example, it may not be obvious whether certain legal search engines employ GenAI to run case queries. Moreover, as customizable GPTs proliferate, companies may begin seamlessly incorporating them into standard system software. And even if lawyers are aware they are using GenAI tools, it may eventually become unreasonable to expect them to verify every GenAI output. While lawyers clearly should not rely on GenAI to perform legal research for them, these tools are useless if lawyers cannot rely — at least to some extent — on their validity.
Finally, courts’ varied approaches to developing AI rules may result in inefficiencies and confusion. In Texas alone, the current procedural requirements for using and disclosing AI are not uniform. For instance, while the Fifth Circuit plans to incorporate its AI policies into its certification of compliance, Judge Starr requires litigants to file a separate form. Courts also require different types of certification. While the proposed Fifth Circuit rule focuses on the accuracyof GenAI-generated content, other courts have honed in on other concerns, like confidentiality. For instance, Judge Stephen Vaden of the United States Court of International Trade recently required litigants to certify that they did not breach client confidentiality by inputting sensitive information into a GenAI program.
These issues make it essential for lawyers to stay informed about developments in GenAI technology and ensure they are complying with applicable court rules. Indeed, the only thing that is certain in this area is that there will continue to be change.
Kaylen E. Strench is an associate in the Appellate Practice Group in Haynes Boone’s Houston office. Before joining the firm, Kaylen clerked for Texas Supreme Court Justice Debra Lehrmann and Fifth Circuit Court of Appeals Judge Catharina Haynes. Her practice focuses on appellate matters in federal and state court.
