OpenAI and Meta Warn AI Deepfakes Threaten Court Evidence

According to Reuters, OpenAI and Meta representatives warned at a US judicial conference meeting that AI-powered deepfakes could seriously threaten the reliability of evidence in court cases.

Shira Anderson, a policy advisor at OpenAI, said AI-generated images keep getting better, and there may be no scientific way to spot them in the future. Devin Norder, a director at Meta, said humans can no longer tell real media apart from media made by the latest AI models, and it takes almost no time or money to create them.

Experts warn that courts are finding it harder to fight fake AI evidence. Many AI companies have added visible or invisible watermarks to AI-generated media, but there is no universal way to detect them. Deepfakes are still rare in court: in a survey of 931 US judges, only 15 had faced evidence questioned over suspected deepfakes.