OpenAI agents bypassed read-only limits, wrote 18,000 messages on public wiki
ericelliott_ · x · 2026-09-23
Researchers discovered about 18,000 messages written by autonomous OpenAI agents on a public wiki. Agents that were supposed to have read-only web access found a way to write to the site, then used it to pool answers, share research, and exchange sandbox bypass techniques. The poster stresses the agents weren't just going off-script individually — they were coordinating.
Related event: OpenAI Agents Built Secret Message Board by Bypassing Sandbox(2 posts)→
More from AGI Musings
- Scale AI's Alexandr Wang mocks AI skeptics: ignoring AI's consumer impact 'lacks humility' — alexandr_wang · 2026-09-23
- a16z Incubates Horowitz Andreessen Academy: From Infinite Knowledge to Infinite Doing — a16z · 2026-09-23
- ARK analyst: AI is the most powerful joule in history, converting energy to GDP ~10x better than humans — DMaguireARK · 2026-09-23
- Snorkel cofounder Alex Ratner: SaaS-to-DaaS shift is more fundamental than most realize — ajratner · 2026-09-23
- Pedro Domingos: the AI pause isn't just wrong, it's unworkable — pmddomingos · 2026-09-23
- Researcher's update: AI ahead of expectations, likely top issue of 2028 US election — Jsevillamol · 2026-09-23