🤖 STOC 2027 Drops Double-Blind and Introduces Six Rules Against AI-Generated Papers
The STOC 2027 program committee — the 59th ACM Symposium on Theory of Computing (Atlanta, June 6–10, 2027, as part of FCRC) — has introduced six experimental rules to counter the influx of AI-generated papers. Double-blind review, in place at STOC since 2023, is being abolished: all listed authors must be human and are responsible for the paper. Mandatory requirements include an arXiv preprint before the deadline (the PDF must match the arXiv version), a 20–30-minute video presentation of the paper, and disclosure of substantial LLM use — minor grammatical edits do not require disclosure. Reviewers have been granted the right to use LLMs when reviewing papers, but only with the explicit consent of the authors; decisions are still made by humans. The submission deadline is November 2, 2026, with a maximum of 5 submissions per author.
🌍 This is the first major theory of computation conference to roll back double-blind review to defend against AI generation, and it restructures the entire peer review cycle: mandatory arXiv preprints close the loophole of 'unfinished' papers, the video verifies that the author understands their own text, and consent to LLM review sets a precedent for two-way AI use. The methodology fundamentally changes: instead of unreliable detection of generated text, the committee verifies the 'text–human' connection, and the 20–30-minute video serves as a liveness check of understanding, not an assessment of quality. If the experiment works, similar rules are likely for ML conferences, where the generative influx arrives first.
👤 If you are submitting: the deadline is November 2, 2026, anonymity is no longer required, and refusing consent to AI review effectively means refusing to submit — consent is a precondition, as confirmed by the PC chair in discussion. The boundary between 'substantial LLM use is disclosed, minor grammatical edits are not' is not operationalized and relies on the honesty of authors. Lance Fortnow warns that video evaluation may harm non-native English-speaking authors.
Source 1: https://blog.computationalcomplexity.org/2026/09/the-new-stoc-rules-for-ai-era.html Source 2: https://acm-stoc.org/stoc2027/stoc2027-cfp.html