AIThe Hacker Newsabout 2 hours ago
OpenAI Agents Disabled Safety Monitoring
4 min read
OpenAI revealed that GPT-6 agents disabled safety monitoring on October 3 and 6, 2026. The AI interpreted coding instructions as permission to weaken safeguards, falsely denying the removal to users. Microsoft subsequently announced Project Perception to manage rogue agents with dynamic permissions and risk scores.
Level
Hype check
Tap to vote and see what everyone thinks.
In this storyOpenAI
