Sanctions Applied for Artificial Intelligence-type Conduct

Once again, a forum has sanctioned a private party for use of non-existent cases and quoting of non-existent dialogue. The Civilian Board of Contract Appeals (“Board”) sanctioned a pro se litigant for its misconduct which possibly resulted from the use of artificial intelligence (“AI”) in drafting his motions. Decision Regarding Sanctions, Louis J. Blazy v. Dept of State, CBCA 7992, 7993, Feb. 24, 2026.

The Board issued appellant seven orders to submit copies of the alleged non-existent decisions and the non-existent deposition, but the appellant failed to supply either, and did not express an apology to the Board, but maintained that the information had been submitted in “good faith” and only resulted in “harmless error.”

The Board refused to excuse the litigant, stating that it “construe[s] a pro se litigant’s pleadings liberally, but such lenience does not affect [the litigant’s] burden of proof or the Board’s assessment of the factual record.” There is a fundamental standard of conduct for all litigants-- the duty of candor to the tribunal, which requires that all submissions be accurate, complete and truthful. Rule 35 of the Board states that if any party fails to comply with Board direction or order, or engages in misconduct affecting the Board, its process, or its proceedings, “the Board may make such orders as are just, including the imposition of appropriate sanctions.”

In sanctioning the appellant, the Board noted:

• Appellant’s submission of non-existent decisions, quoting non-existent deposition language and mischaracterizing existing deposition languages violated appellant’s duty of candor;

• Appellant’s actions violated Board rule 35, wasted the Board’s and opposing counsel’s resources, and increased the time for deliberation; and

• Appellant’s vague, confusing the unpersuasive responses to the Board’s orders raise questions as to the credibility and veracity of its responses.

In a footnote, the Board stated that the appellant stated that he did not use AI in preparing his motion. The Board also stated that “[a] characteristic of using AI is its propensity to ‘hallucinate’ meaning to create information which appears credible and real but which is actually false, non-existing or a mischaracterization of existing information.” Although the use of AI is not prohibited at the Board, “[i]f using AI, parties retain full responsibility for the accuracy of all submissions to the Board.” CBCA Annual Report Fiscal Year 2025, Feb. 23, 2026.

The Board held that the appellant’s actions constituted sanctionable conduct pursuant to Board Rule 35, and formally admonished the appellant that if he continued such conduct, additional and harsher sanction will be imposed, which could include dismissal of the pending appeal without proceeding to the merits.

For other helpful suggestions on government contracting, visit:

Richard D. Lieberman’s FAR Consulting & Training at https://www.richarddlieberman.com/, and Mistakes in Government Contracting at https://richarddlieberman.wixsite.com/mistakes