US Judiciary Panel Forms Subcommittee on AI Deepfake Evidence After OpenAI, Meta Testimony

On October 15, 2026, representatives from OpenAI and Meta Platforms told the US Judicial Conference's Advisory Committee on Evidence Rules, meeting in Boston, that AI-generated deepfakes are becoming so sophisticated that verifying the authenticity of audio and video evidence at trial will grow more difficult. The committee's chair responded by appointing a subcommittee to continue examining the issue, extending a process the panel has run since 2023.

What it is

The Advisory Committee on Evidence Rules is a federal judiciary rulemaking body that has been researching for three years whether the existing Evidence Rules adequately handle AI-generated and AI-altered material. Two draft measures are under study: proposed Rule 707, which would apply expert-testimony reliability standards to machine-generated evidence, and proposed Rule 901(c), aimed at authenticating deepfakes. At its May 7, 2026 meeting the committee did not recommend action on proposed Rule 707, instead revising it for further study alongside the broader deepfakes issue.

What it requires

No AI-specific evidence rule is currently in force; both proposals remain study items rather than approved text. As drafted, proposed Rule 901(c) would require a challenger to present evidence sufficient to support a finding of fabrication before a court inquires further into authenticity. The public comment period on both proposals closed February 16, 2026. A Federal Judicial Center survey released in March 2026 found only 15 of 931 responding judges had encountered a case where evidence was challenged as a deepfake, a data point the committee is weighing against calls for new rules. OpenAI's policy counsel told the committee that each company uses its own watermarking system and there is no universal means to verify the authenticity of visual content.

What security leaders should do

Security and legal teams handling litigation involving audio, video or synthetic media should not wait for Rule 707 or 901(c) to take effect, since neither has final approval and the often-cited December 1, 2027 effective date is not locked in. Build authentication and chain-of-custody practices now under the evidence rules already on the books, since courts are already excluding AI-altered material and requiring disclosure of AI-generated exhibits. Track the committee's next meeting output given the newly appointed subcommittee, and treat any internal use of watermarking or provenance tools as one input among several rather than a reliable sole safeguard, since providers themselves say no universal detection standard exists.