Federal judge chooses not to sanction lawyer who admitted using AI in mistake-filled brief

In a significant ruling that highlights the growing intersection of technology and the legal profession, U.S. District Judge Thomas Cullen recently declined to impose sanctions on attorney Thomas Guyer, despite the submission of a brief containing incorrect case citations and quotes generated by artificial intelligence (AI). The decision underscores the challenges and ethical considerations lawyers face as they increasingly rely on generative AI tools for legal research and drafting. While Judge Cullen acknowledged Guyer’s mistake, he noted that the attorney took full responsibility, demonstrating remorse and a commitment to ensuring such errors would not recur in the future. The case serves as a reminder of both the potential benefits and pitfalls of AI in the legal field, as well as the need for attorneys to remain vigilant in upholding professional standards.

The incident began when Guyer, a seasoned and respected attorney with a long-standing reputation for excellence, submitted a brief in a case involving his client, Karen Iovino, who alleged she was fired for reporting concerns about her employer’s contract with the State Department. Unbeknownst to Guyer, the brief contained inaccuracies—specifically, misquoted and miscited cases—that were generated by an AI tool, Claude 3 Opus, developed by Atrophic Inc. Guyer later explained that he utilizes a suite of AI technologies, including GPT, for legal research and document preparation, describing these tools as capable of producing “excellent to brilliant legal arguments.” However, in this instance, the AI tool fell short, leading to what Guyer characterized as “fictitious” citations and misquotes, which he only discovered after the document was filed.

During an October 2024 hearing, Judge Cullen addressed the matter, emphasizing that while Guyer’s actions were not intentional, they reflected one of the inherent risks of relying on generative AI. The judge noted that Guyer, who has an “unblemished record,” promptly owned up to the mistake and took sole responsibility, actions that Cullen viewed favorably. The judge also highlighted Guyer’s transparency, pointing out that the attorney had identified errors missed not only by opposing counsel but also by the court itself. This level of accountability, Cullen suggested, demonstrated Guyer’s commitment to ethical practice and his understanding of the responsibilities that come with using advanced technologies in legal work.

The broader implications of this case extend far beyond Guyer’s individual actions, as it touches on a critical issue facing the legal profession: the integration of AI into legal practice and the potential for errors or misleading information. Judge Cullen observed that the use of generative AI is becoming the “new normal” in the legal field, with many attorneys turning to these tools to streamline research, drafting, and other tasks. However, he emphasized that the adoption of such technologies does not absolve attorneys of their professional obligations. “Lawyers who use generative AI must still adhere to the basic tenets of conduct,” Cullen stated, including the duty to take reasonable measures to ensure the accuracy of filings. The case serves as a cautionary tale, reminding lawyers that while AI can be a powerful tool, it is not a substitute for careful oversight and scrutiny.

In the aftermath of the incident, both the Virginia State Bar and the Oregon State Bar (where Guyer is licensed) opened investigations into the matter. Guyer, demonstrating his commitment to accountability, proactively reported himself to the Oregon bar. While Judge Cullen expressed hope that his decision would provide guidance for these investigations, no formal developments or conclusions have been announced by either bar association as of yet. The outcome of these investigations will likely set an important precedent for how state bars address the challenges posed by AI in legal practice, particularly as more attorneys begin to use these tools in their work.

Guyer’s attorney, Denis Quinn, echoed the judge’s sentiments, describing his client as “incredibly remorseful” for the mistake. Quinn emphasized that Guyer’s regret was not merely a defensive response but a genuine reflection of his commitment to upholding the integrity of the legal profession. “This incident will ensure that he doesn’t do that again,” Quinn said, suggesting that the experience would serve as a lasting lesson for Guyer and, by extension, for other attorneys considering the use of AI in their work.

Ultimately, this case highlights the double-edged nature of generative AI in the legal field. On one hand, these tools offer unprecedented opportunities to enhance efficiency and innovation, enabling attorneys to craft compelling arguments and conduct complex research with greater speed and accuracy. On the other hand, the potential for errors—or even the creation of entirely fictitious information—poses significant risks, both to the integrity of legal proceedings and to the reputation of the profession as a whole. As Judge Cullen’s decision makes clear, the onus remains on attorneys to ensure that the benefits of AI are realized without compromising the ethical standards that underpin the legal system. The outcome of this case, and the subsequent actions of state bar associations, will likely shape the future of AI’s role in the courts, offering both a roadmap for responsible use and a reminder of the enduring importance of human oversight.