Federal Judge Mandates AI Disclosure in All Court Filings, Citing ‘Erosion of Trust’

Federal Judge Mandates AI Disclosure in All Court Filings, Citing ‘Erosion of Trust’

A federal district judge in the Southern District of New York issued a sweeping standing order Wednesday requiring every attorney filing a document in her courtroom to disclose whether generative artificial intelligence was used in its preparation, marking the strictest judicial response yet to the rapid adoption of AI tools in litigation.

Judge Sarah K. Timmons’s order applies to all motions, briefs, and exhibits submitted in civil and criminal cases before her. Attorneys must file a certification stating which AI system was used, whether the output was reviewed by a human, and that the filing is “factually accurate and not hallucinated.” The order takes effect July 1, 2026. “The proliferation of unverified AI-generated content is eroding the trust that underpins our adversarial system,” Timmons wrote. “This court will not serve as a beta test for unregulated language models.”

Legal ethics experts said the order is likely to spark a wave of similar mandates. “Judges are scrambling to catch up,” said Rebecca Lin, a professor of legal ethics at Georgetown University. “What started as a few isolated cases of hallucinated citations has become a systemic concern. Timmons’s order is a shot across the bow.” The Standing Committee on Professional Responsibility of the American Bar Association is reportedly drafting a model rule requiring disclosure of AI use, though it has not yet been released.

Critics argue the order could burden small firms and solo practitioners who rely on affordable AI tools for legal research and drafting. “This creates a two-tiered system,” said defense attorney Marcus Davila. “Big firms already have in-house compliance teams. Small shops will face chilling-effect audits.” The order exempts routine clerical functions such as spell-checking and calendaring. Practitioners should expect additional judicial guidance before the effective date, including a potential pilot program for electronic filing metadata tags that automatically flag AI involvement.