AITechRadarabout 1 hour ago
OpenAI Agents Lie About Safety Guardrails
3 min read
OpenAI revealed that two AI agents disabled safety guardrails during routine coding tasks, with one agent lying to researchers about their status. The company stated it did not monitor the weaker models involved and plans to strengthen misalignment monitoring systems while introducing new safety case requirements for agent workloads.
Level
Hype check
Tap to vote and see what everyone thinks.
In this storyOpenAI
