AI news · October 17, 2026
OpenAI and Meta warn judges that deepfakes may outrun detection
OpenAI and Meta told a US judicial panel that deepfake detection may not scale to every case; evidence handling will need source records, expert review, and more than one signal.
The Neuron reported on October 16 that OpenAI and Meta warned a federal judicial panel that synthetic audio and video are becoming difficult to distinguish from genuine evidence. OpenAI policy counsel Shira Anderson said she did not expect a scientific method that can reliably identify every AI-generated image. The warning is about court evidence, not a new detector product.
That leaves a practical burden on the people handling a file: preserve the original, record how it moved, compare it with independent material, and document expert judgment. A detector can help, but a positive or negative result cannot carry the whole decision. The warning also matters for builders of media tools because a polished output may be challenged after it leaves the product.
What you can do with it
Build an evidence record beside the generated or uploaded file: original bytes, hash, source, timestamps, transformations, model settings, detector results, and human decision. Test the record on a disputed clip and make sure a reviewer can reconstruct what happened without trusting one classifier.
Our take
The important shift is from detector confidence to evidence handling. A single score is easy to display and hard to defend when media changes or a new generator appears. Products that keep the source and review trail will be more useful than products that only claim to identify synthetic media.
Source: The Neuron — Made With Models writes the brief; the reporting is theirs.