OpenAI and Meta warn US judges that deepfakes may be undetectable

At a meeting of the US Judicial Conference’s Advisory Committee on Evidence Rules in Boston, an OpenAI policy counsel said she does not expect any…

October 15, 2026 · Government

At a meeting of the US Judicial Conference’s Advisory Committee on Evidence Rules in Boston, an OpenAI policy counsel said she does not expect any scientific method to reliably detect AI-made images in future. A Meta product director said people can no longer tell recent AI media from the real thing. Both said watermarks help but will never be fully reliable, partly because each company uses its own system. The committee has been weighing rule changes on deepfake evidence since 2023; a March survey found 15 of 931 judges had faced a deepfake challenge.

Why it matters to you: If experts cannot reliably spot fakes, checking where a photo, video or voice clip came from matters more than how real it looks, at work, at school and in court.

Try this: Build a habit for checking a suspicious clip before you share it.

Give me a 5-step checklist for deciding whether a video, photo or voice clip might be AI-generated before I share it or act on it, including how to trace the original source.

Source: Reuters via Investing.com

More from Today in AI

All of Today in AI. Where AI is heading.

Related Power of AI pages

Keep reading with Start here, everyday uses, the tool guide, the writing workshop, and sources and standards.