[D] Dealing with an unprofessional reviewer using fake references and personal attacks in ICML26

An anonymous author has sparked a major discussion in the AI research community by detailing an allegedly fraudulent and hostile peer review experience for the International Conference on Machine Learning (ICML) 2026. The author claims a reviewer gave their paper the lowest possible score (1 out of 5) with high confidence, while two other reviewers awarded the highest score (5). The core allegation is that this reviewer relied on "fake references"—citations that do not exist or are irrelevant—and employed personal attacks, calling the authors "close-minded" and "hostile." The reviewer's critique is further described as containing "mathematically nonsensical proofs" and baseless accusations about licensing violations.

The author describes bizarre behavioral patterns, including the reviewer constantly editing a "PS" section of their review to bait the Program Chair and using aggressive formatting with syntax errors like bolding that ends with periods (**.). Despite professional rebuttals that addressed weaknesses and later pointed out the reviewer's irrelevant references and circular reasoning, the reviewer allegedly ignored the response. The post is a call for advice from the research community on whether such a review can be formally discarded or flagged for intervention by an Area Chair (AC), highlighting a critical vulnerability in the peer-review system when a single actor operates in bad faith.