More than a fifth of reviewers used AI after being told not to.
The machine learning conference ICML split its 17,000 reviewers into two groups this year. One group was banned from using chatbots, and 22.5% of them reported using one anyway. Paper scores and accept decisions came out almost identical in both groups.
Why it mattersPeer review decides which science gets believed, and the rules written for chatbots are not holding. The experiment also suggests the tools changed how reviews read, not who got in.
Read in the app — free, in 9 languages
Related stories
Microsoft ran 1,024 agents on one job with nobody in charge.2026-09-29An AI reran 4,452 economics papers and flagged most of them.
2026-09-29A wrong AI summary cut what people remembered almost in half.
2026-09-29Britain's testers saw Astra attack code nobody asked it to touch.
2026-09-29OpenAI backs a lab making drugs for viruses that do not exist.
2026-09-28