Federal evidence panel keeps AI risks under review as deepfakes improve

On October 15, 2026, the U.S. Judicial Conference’s Advisory Committee on Evidence Rules met on AI-generated evidence, and Reuters reported that representatives from OpenAI and Meta warned that synthetic audio and video are becoming increasingly difficult to distinguish from authentic media. The committee appointed a subcommittee for further work, while its published agenda materials argued that existing Rules 702 and 703 can screen AI-assisted expert evidence for reliability; only 15 of 931 judges in a judiciary survey had encountered a deepfake-evidence challenge. Litigation teams should preserve provenance, tool versions, prompts, transformation logs and expert-validation records now, even if no AI-specific evidence rule is adopted. The operational risk is less a universal admissibility ban than a fact-intensive challenge to reliability that may be impossible to reconstruct after a matter develops.