ICLR AI review detector saga: no false positives, flagged reviews were just visible to all
guyvdb · x · 2026-11-18
guyvdb investigated the ICLR AI-detector controversy and found the 36-year friend of RaoZoghzoghi was NOT flagged — their reviews scored near 0% AI. What actually happened: other reviewers on the same paper were flagged as AI, and all reviewers can see this info (and receive an email about it). The system produced no false positives; the issue is the visibility of flags.
More from AGI Musings
- Software is a constitution: the case for codified rules in human-agent collaboration — danshipper · 2026-11-18
- a16z podcast: OpenAI's Mark Chen and John Schulman on recursive self-improvement and alignment — a16z · 2026-11-18
- A Running List of Prominent Figures Calling to Pause Further AI Development — erikphoel · 2026-11-18
- Report: enterprise spend on agent data access to hit ~$30B by 2030 — BenBajarin · 2026-11-18
- NSA outage was just a software update; Microsoft incident traced to 4-month-old latent bug — TheZvi · 2026-11-18
- Matt Levine on the antitrust paradox: insurance loophole for AI safety coordination — zacharynado · 2026-11-18