The Importance of Verifying Your Use of AI for Litigation

You are handing a case involving millions of pages of documentation, emails, etc., including documentation with sensitive trade secrets and intellectual property.  You are under the gun to submit a brief in opposition to a motion seeking injunctive relief.  You want to use state of the art tools for jury selection.  You want to do predictive modeling for an analysis of whether your client should settle a case and for how much.  You might think the answer to these tasks is artificial intelligence (“AI”).  Not so fast…and, more importantly, not without reviewing, verifying and using the skills, instincts and experience of human professionals. No, lawyers are not becoming obsolete and AI, in its current iteration, is showing how valuable human intelligence and experience is.

The pitfalls of blindly using AI, in lieu of your good old legal skills and analysis, have been most glaring and prominent in legal research and writing.  Most attorneys now know how to open the internet or an AI app and ask for answers to a legal question, and many of attorneys may have already tried using the results of those searches to draft a brief or other argument.  Some attorneys, albeit a fewer number, however, have done so blindly, trusting, but not verifying, that what AI generated was accurate, going so far as to cite and quote those cases.  Those attorneys did so at their own peril and to the detriment of their reputations, wallets, and clients’ cases.

In Federal Court, Rule 11(b)(2) mandates attorneys ensure “the claims, defenses, and other contentions are warranted by existing law or by a nonfrivolous argument for extending, modifying, or reversing existing law or for establishing new law.”  See F.R.C.P. 11(b)(2).  “The central purpose of Rule 11 is to deter baseless filings in district court and thus…streamline the administration and procedure of the federal courts.”  Cordova v. Univ. Hosp. & Clinics, Inc., 92 F. 4th 266, 273 (5th Cir. 2024) (quoting Cooter & Gell v. Hartmarx Corp., 496 U.S. 384, 393 (1990)). Courts apply an “objective, not subjective, standard of reasonableness.”  Snow Ingredients, Inc. v. Snowizard, Inc., 833 F.3d 512, 528 (5th Cir. 2016) (citing Whitehead v. Food Max of Miss., Inc., 332 F.3d 796, 802 (5th Cir. 2003)).  “At the very least, the duties imposed by Rule 11 require that attorneys read, and thereby confirm the existence and validity of, the legal authorities on which they rely.”  Park v. Kim, 91 F. 4th 610, 615 (2d Cir. 2024); see also Mata v. Avianca, Inc., 678 F. Supp. 3d 443, 448 (S.D.N.Y. 2023).

Unfortunately, AI searches have, at times, generated “hallucinated” cases, meaning that the citations and/or the proposition that the case stands for are not accurate, and in some instances, are not real cases.  These citations may even look and sound like real decisions. Worse yet, attorneys have cited these hallucinated cases in legal filings and in arguments before courts throughout the country. 

Federal and state courts throughout Pennsylvania, New Jersey and around the country have taken an array of measures to either ensure the accuracy of submissions or bar the use of AI completely.   One of the earliest adopters of the “trust but verify” approach taken by courts came from The Honorable Michael Baylson, in the United States District Court for the Eastern District of Pennsylvania, who in 2023 issued the following standing order…

STANDING ORDER RE: ARTIFICIAL INTELLIGENCE (“AI”) IN CASES ASSIGNED TO JUDGE BAYLSON – “If any attorney for a party, or a pro se party, has used Artificial Intelligence (“AI”) in the preparation of any complaint, answer, motion, brief, or other paper, filed with the Court, and assigned to Judge Michael M. Baylson, MUST, in a clear and plain factual statement, disclose that AI has been used in any way in the preparation of the filing, and CERTIFY, that each and every citation to the law or the record in the paper, has been verified as accurate.”

Other judges in the Eastern and Middle Districts of Pennsylvania have followed suit.  The same holds true in the District of New Jersey, where Judge Evelyn Padin added the following to her General Pretrial and Trial Procedures…

1(b) “The use of any GAI (e.g., OpenAI’s ChatGPT or Google’s Bard) in any court filings requires a mandatory disclosure/certification that: (1) identifies the GAI program; (2) identifies the portion of the filing drafted by GAI; and (3) certifies that the GAI work product was diligently reviewed by a human being for accuracy and applicability.

The Eastern District of Texas standardized their local rules to address this concern as well, by amending the local rules as follows:

“If the lawyer, in the exercise of his or her professional legal judgment, believes that the client is best served by the use of technology (e.g., ChatGPT, Google Bard, Bing AI Chat, or generative artificial intelligence services), then the lawyer is cautioned that certain technologies may produce factually or legally inaccurate content and should never replace the lawyer’s most important asset – the exercise of independent legal judgment. If a lawyer chooses to employ technology in representing a client, the lawyer continues to be bound by the requirements of Federal Rule of Civil Procedure 11, Local Rule AT-3, and all other applicable standards of practice and must review and verify any computer-generated content to ensure that it complies with all such standards.” Local Rule AT-3(m).

While most courts have adopted the “trust but verify” approach to using AI as a tool, some courts have been less tolerant of the use of this generation of AI.  In fact, the United States District Court of the Western District of North Carolina (Charlotte Division) issued a standing order barring the use of most AI research, except for those embedded in standard online research sources.

“All attorneys and pro se filers must file a certification with any brief or memorandum submitted to the Court that states the following:

• No artificial intelligence was employed in doing the research for the preparation of this document, with the exception of such artificial intelligence embedded in the standard on-line research sources WestLaw, FastCase, and Bloomberg;
• Every statement and every citation to an authority contained in this document has been checked by an attorney in this case and/or paralegal working at his/her direction (or the party making the filing if acting pro se) as to the accuracy of the proposition for which it is offered, and the citation to authority provided.”

Understandably, courts have been concerned by the unchecked/unverified use of AI in court submissions.  And, when that lack of human verification results in the submission of” hallucinated” cases that either don’t stand for the proposition submitted or don’t exist at all, courts have been clear about their frustration, resulting in stern opinions and a variety of sanctions.  In the Eastern District of Texas, a judge rebuked an attorney who “failed to make a reasonable inquiry required by Rule 11 and long-standing precedent into the validity of the arguments he presented.” See Gauthier v. Goodyear Tire & Rubber Co., No. 1:23-cv-00281 2024 U.S. Dist. LEXIS 214029 at *7 (E.D. Tex. Nov. 25, 2024).  As a result of their failure, the attorneys were sanctioned in the amount of two thousand dollars and ordered to attend a CLE on generative AI in the legal field and submit proof of attendance.

In the Southern District of New York, the court went further, explaining the multitude of harms that result from an attorney’s failure to use AI unchecked. 

“Many harms flow from the submission of fake opinions. The opposing party wastes time and money in exposing the deception.  The Court’s time is taken from other important endeavors. The client may be deprived of arguments based on authentic judicial precedents. There is potential harm to the reputation of judges and courts whose names are falsely invoked as authors of the bogus opinions and to the reputation of a party attributed with fictional conduct. It promotes cynicism about the legal profession and the American judicial system.  And a future litigant may be tempted to defy a judicial ruling by disingenuously claiming doubt about its authenticity.”

Mata v. Avianca, Inc., 678 F. Supp. 3d 443, 448-49 (S.D.N.Y 2023).

The court in Mata issued five thousand dollars in monetary sanctions and required the attorney to send a “letter individually addressed to each judge falsely identified as the author of the fake … opinions.”  The letter was required to include a copy of the Court’s sanction order and opinion and a copy of the “’fake’ opinion attributed to the recipient judge.”

Courts and the rules of professional conduct make it clear to use AI wisely and only as a supplement to the skills and instincts that you, as a human and professional, have honed throughout your career.  And when you do use it, understand that you will be required to verify every word and every case…and even certify the same to the court. Failure to do your due diligence can result in severe consequences to attorneys and their clients.