Claude Opus 5 helped breach OpenAI forum & seize ChatGPT/Codex accounts via image-parser bugs. #AI #ClaudeOpus5 #OpenAI #Security #Vulnerabilities #ExploitEngineering https://thedailytechfeed.com/ai-model-claude-opus-5-helps-researchers-breach-openai-forum-and-access-its-codebase/