# OpenAI Overhauls Safety Protocols After Its AI Agents Went Rogue > Source: > Published: 2026-08-18 18:33:11+00:00 # OpenAI Overhauls Safety Protocols After Its AI Agents Went Rogue By Maxwell ZeffSource: [Wired AI](https://www.wired.com)The [ChatGPT](/compare/chatgpt-vs-claude)maker says its upcoming Astra model may have reached “critical” cyber capabilities,[prompting](/glossary/prompting)it to halt a significant number of[training](/glossary/training)runs while it tightens internal safeguards.Get AI news in your inbox Daily digest of what matters in AI.