Post #1095981

ICML'26 had two groups of reviewers, one for which LLM use was forbidden and another for which it was not. Reviewers could state their preference and only those who claimed to be ok with forbidding LLM use were assigned to that group. But then, ~500 of the reviewers in the LLM-forbidden group were caught using LLMs through hidden watermarks in the papers they reviewed. As a consequence, ~400 of those reviewers had their submissions desk-rejected (~500 rejections). https://blog.icml.cc/2026/03/18/on-violations-of-llm-review-policies/ via https://retractionwatch.com/2026/03/28/weekend-reads-illicit-ai-use-peer-reviews-commentary-talc-retracted-coauthorship-traded-commodity/