ICLR 2027 caps authors at 20 papers, mandates AI-use statement

The ICLR 2027 author guidelines tighten the screws on what has become an unmanageable flood of machine-learning submissions. No author may appear as a co-author on more than 20 papers this cycle, every submission must have at least one author who is registered to review at least three papers, and authors whose names appear on three or more submissions are expected to review at least six of their own.

The submission mechanics are worth planning around. Main text should be 9 pages or fewer at submission and is increased to 10 pages during discussion and rebuttal, with unlimited references and appendices on top. An AI use statement is required and does not count toward the page limit, though the specifics of what to disclose sit in a separate 'AI Policy for Authors' document that the guidelines page points at rather than reproducing.

The calendar is tight. Abstracts are due September 18, 2026 AOE, full papers a week later on September 25, reviews return October 29 and decisions are announced December 9. Double-blind is being enforced without mercy: 'any paper where author identity is revealed in either the main text or the supplementary material will be desk rejected.' Preprints on non-peer-reviewed sites like arXiv are still permitted if referenced in the third person.

Two of the sharper details are the withdrawal and dual-submission rules. Papers pulled after the deadline remain hosted by OpenReview in a publicly visible 'withdrawn papers' section and are de-anonymized immediately, which changes the calculus for speculative or borderline work. Anything 'identical (or substantially similar) to versions that have been previously published' is out.

The honest caveat is that the AI policy specifics and the program-chair roster are not on this page, so the rules most likely to shape how LLM-assisted writing is treated are still to come. What the mechanics do tell you is where the conference wants the pressure to land, which is on the largest labs and on reviewers who have been coasting. If you run a group, plan your 2026 submission slate and your reviewing commitments in the same conversation, not two separate ones.