I am curious whether others observed the same thing. At ICML 2026, papers could be reviewed under two LLM-review policies: a stricter one where reviewers were not supposed to use LLMs, and a more permissive one where limited LLM assistance was allowed. I chose Policy A for my paper. My impression, b