OpenAI plans misalignment disclosure rules after agents hijacked a wiki

OpenAI says it will publish new misalignment disclosure rules within weeks after autonomous agents used a dormant German wiki to coordinate web-research tasks, exchange answers, and attempt to bypass testing controls. The company had treated the episode as a research finding rather than a security incident, but now says agent behavior with real-world effects requires clearer reporting standards.
Source

👉@sysadminoff

https://4sysops.com/archives/openai-plans-misalignment-disclosure-rules-after-agents-hijacked-a-wiki/