OpenAI and Meta Warn Federal Judges That Deepfake Evidence Will Be Harder to Verify

OpenAI and Meta representatives told the US Judicial Conference's Advisory Committee on Evidence Rules that AI deepfakes are becoming hard to distinguish from authentic material.

What Happened

The committee met on October 15, 2026, in Boston.^[2] According to Investing.com, the representatives warned that verifying the authenticity of audio and video evidence in court will grow more difficult.^[1] The agenda for the meeting states: "At this meeting, the Committee has convened a panel of experts to discuss the problem of deepfakes and to provide suggestions about a proposed rule."^[2]

Shira Anderson is policy counsel at OpenAI. According to Investing.com, she said AI images keep improving. She said she does not expect a scientific method to detect them in the future.^[1] She also said each company uses its own watermarking and verification system. So there is no universal way to detect all watermarks, and these systems will never be completely reliable.^[1]

Devin Norder is director of product management at Meta. He said humans cannot distinguish media created by recent AI models from authentic content. He added that making it takes minimal cost and time.^[1]

The committee is weighing a draft rule on items challenged as deepfakes. The agenda book says: "The Committee has prepared, but not finally approved, a new Rule 901(c)."^[2] It also notes that "Judge Furman summarized the Committee's work on AI that began in 2023."^[2]

Why It Matters

So far, deepfake disputes are rare in federal courts. The Federal Judicial Center surveyed all federal trial court judges. "Of the 931 judges who responded, only 15 reported having encountered a deepfake challenge."^[2] U.S. District Judge Jesse Furman chairs the panel. He noted that the issue has not yet appeared often in actual cases.^[1]

The concern is what comes next. Meta's Norder told the committee that humans cannot tell media from recent AI models apart from authentic content. OpenAI's Anderson said watermarking systems will never be completely reliable.^[1] New laws in California and the European Union require watermarking to detect deepfakes, Investing.com reports. But each company uses its own system.^[1]

What to Do

You may be a lawyer, an investigator, a claims handler, or anyone handed a photo or video as proof. A few habits help.

• Keep the original. Ask for the original file, not a screenshot or a forwarded copy. Preserve its metadata.
• Ask where it came from. Who captured it, on what device, and how did it reach you? A clear chain of custody matters as much as the file itself.
• Do not rely on a missing watermark. Watermarking systems differ between companies. The absence of a label does not mean a file is real.
• Get a second pair of eyes. Use a detection tool as one input alongside other evidence. No single check is final.

Verify What You See and Hear

Upload any suspicious image or video frame at witness.vision/scan. Witness checks whether it is real or synthetic. The Witness Chrome extension scans images directly on any webpage.

Scan now at witness.vision/scan

Related reading

• C2PA Content Credentials Missing? What It Actually Tells You
• Deepfakes Are Now a Compliance Problem: What Changed in 2026
• Why Your Eyes Can't Spot Deepfakes