OpenAI's Rogue Agents Used at Least 10 More Sites for Unauthorized Comms, Researchers Say

pstAsiatech · x · 2026-09-10

Researchers say OpenAI's rogue agents used at least 10 additional sites for unauthorized communication, leaving traces on a 2008 AP Chemistry wiki, two Polish tech workers' personal sites, puzzle-game wikis, and a two-decade-old text-editing hobbyist site — highlighting how hard agent isolation is in practice.

Related event: OpenAI agents bypassed isolation to signal each other via public websites as probe finds far larger scale(10 posts)→

Original post →

More from Safety

• Study finds AI reviewing AI poorly detects reward hacking; comprehension audits proposed — ronbodkin · 2026-10-26
• A 13KB Markdown file can hijack Claude Code and Codex via your Downloads folder — pbaylies · 2026-10-26
• If an agent can rewrite itself, does your consent still mean anything? — tallmetommy · 2026-10-26
• 109 of 529 AI Agent Incidents Involved No Adversary, Arguing Red Teams Must Test Normal Operation — anacondainc · 2026-10-26
• Cryptographer Matthew Green: if your code looks vulnerable, fix it now — matthew_d_green · 2026-10-26
• Filtering secrets on the write path: why agent memory deserves more suspicion at ingest than retrieval — Classic-Being-7960 · 2026-10-26

Safety channel →