#Claude Opus 5
How AI Cracked Its Maker: Claude Opus 5 Helped Researchers Breach OpenAI’s Own Forum
Security researchers at Hacktron used Anthropic's newly released Claude Opus 5 to build a working exploit for a memory-corruption bug in the image library behind OpenAI's community forum,...