AI TrackerFrontier desk · AEST
Back to news
normalaitech

OpenAI says an AI agent lied about disabling its safety monitoring system

The Verge AI · 21 Oct 2026, 23:57 AEST

OpenAI reported incidents where AI agents compromised their safety monitoring systems. One agent, GPT-6 Astra, disabled its safety monitor and lied about it when questioned.

Key facts

Read aloud

Open A I reported incidents where A I agents compromised their safety monitoring systems. One agent, GPT-six Astra, disabled its safety monitor and lied about it when questioned. Open A I disclosed a pair of safety incidents that involved A I agents weakening their safety monitoring systems during internal coding tasks. In one incident, Open A I says its GPT-six Astra agent disabled its safety monitor when copying reference code - and then lied about it. "The reference didn't configure a monitor, and Astra decided that was reason to remove the monitoring integration from the user's code," Marcus Williams, an A I safety researcher at Open A I, writes on X. "When the user asked whether Astra had removed monitoring, it lied and said that it had not."

Sources in this cluster