OpenAI says an AI agent lied about disabling its safety monitoring system
The Verge AIen
OpenAI disclosed a pair of safety incidents that involved AI agents weakening their safety monitoring systems during internal coding tasks. In one incident, OpenAI says its GPT-6 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 […]
This is a short summary published by AI Global Wire. The full article is owned and hosted by The Verge AI — open it there to read it in full.
Read the full story at The Verge AI- OpenAI
- Agenter
Related AI news
- Inception's Mercury 3 diffusion model claims to beat GPT-6 Luna and Claude Haiku 5.5 at a fraction of the waitThe Decoder · October 21, 2026
- OpenAI says the PHP flaws, found using GPT-5.6 Sol, "are among several significant vulnerabilities" it found and expects "similar findings to increase sharply" (OpenAI)Techmeme · October 21, 2026
- Cloudflare says it deployed default WAF mitigations for all free and paid tier customers to combat two PHP flaws, roughly nine hours ahead of public disclosure (Cloudflare)Techmeme · October 21, 2026
- Cloudflare says it deployed WAF mitigations for all free and paid tier customers to combat two PHP flaws by default, about nine hours ahead of public disclosure (Cloudflare)Techmeme · October 21, 2026
- PHP discloses a critical memory-management vulnerability, dubbed "PHP2Shell", that affects up to 70% of websites, discovered using GPT-5.6 Sol in August (Robert McMillan/Wall Street Journal)Techmeme · October 21, 2026
- PHP's creators disclose a critical memory-management vulnerability, dubbed "PHP2Shell", that affects up to 70% of websites, discovered via GPT-5.6 Sol in August (Robert McMillan/Wall Street Journal)Techmeme · October 21, 2026