“Confirm You’re Not a Robot”: AI-Written Briefs Could Lead to Sanctions

On May 7, 2025, GAO issued a decision in Raven Investigations & Security Consulting, LLC, B-423447, warning the bid protest bar that artificial intelligence (“AI”)-based tools utilized without proper oversight may result in severe consequences, including dismissal of the protest and sanctions.

In Raven Investigations, GAO noted numerous inconsistencies in a pro se protester’s filing, including:

• purported direct quotations from GAO that could not be traced back to cited GAO decisions;
• citations to purported GAO decisions that could be located via B-number or Comptroller General’s Procurement Decisions (“CPD”) citations but did not support the principles for which the protester cited them; and
• citations to purported GAO decisions that could not be found.

When GAO questioned these citations, the protester admitted the identified irregularities resulted in part from its use of AI-assisted tools. GAO found that the protester’s explanation failed to excuse the use of improper citations, noting the Court of Federal Claims’ recent decision in Sanders v. United States, No. 24-cv-1301, 2025 WL 957666 (Fed. Cl. Mar. 31, 2025), which cautioned that “the use of AI programs to draft or assist in drafting legal briefs can—and seemingly often does—result in the citation of non‑existent cases.” GAO further explained that, “to the extent the protester used AI tools to help draft its responses . . . without engaging in any review of the material for accuracy . . . that practice wastes the time of all parties and GAO, and is at odds with the statutory mandate that our bid protest forum provide for ‘the inexpensive and expeditious resolution of protests.’”

Though GAO ultimately declined to impose sanctions on the protester, it cautioned pro se litigants and represented parties alike that GAO may impose sanctions against a protester whose actions undermine the integrity and effectiveness of the GAO process, referencing the Supreme Court’s instruction that a forum’s ability to levy sanctions in the face of “abusive litigation practices” as “ancient in origin.” For attorneys in particular, GAO noted that reliance on AI-generated tools without proper validation could violate professional rules of conduct and GAO protective order obligations, thus potentially resulting in attorney discipline.