OpenAI and Meta Issue Joint Warning: Deepfakes Will Severely Undermine the Reliability of Evidence in Legal Proceedings

On October 20, according to Reuters, representatives from OpenAI and Meta both stated at the U.S. Judicial Conference's Advisory Committee on Evidence Rules meeting held on the 15th of this month local time that AI-driven deepfakes will severely threaten the reliability of evidence in legal proceedings.

Shira Anderson, policy advisor at OpenAI, stated: The quality of AI-generated images is continuously improving, and I believe it will be impossible to identify them through scientific means in the future. Devin Norder, Director of Product Management at Meta, noted: Humans can no longer distinguish between authentic media and media generated by at least the most recent generations of cutting-edge AI models. Moreover, the cost and time required to produce such media are currently negligible.

Experts warn that the difficulty for courts in resisting fabricated evidence generated by artificial intelligence is increasing daily. Currently, multiple AI companies have added visible or invisible watermarks to AI-generated media, but there is still no universal method for detecting these watermarks.

However, deepfakes remain uncommon in judicial practice: in a survey involving 931 American judges, only 15 had experienced cases where evidence credibility was challenged due to suspected deepfakes.