ICML Desk Rejects 2% of Submitted Papers for Using LLMs in Peer Review

ICML Blog·medium signal

The International Conference on Machine Learning disclosed it desk-rejected approximately 2% of submitted papers after detecting that authors had used LLMs to write their peer reviews — a direct violation of ICML's review integrity policies. The disclosure is one of the first concrete enforcement actions by a top-tier ML venue against LLM-assisted reviewing and signals that academic publishing is moving from policy statements to active enforcement. The HN post received 33 points and 13 comments.