🚨HUGE: OpenAI was reportedly breached using Anthropic’s AI models, per FT. Researchers at Hacktron AI used Claude to exploit a vulnerability in a third-party forum used by OpenAI. That access extended to multiple OpenAI employees’ ChatGPT accounts, ultimately giving researchers a path into OpenAI’s internal GitHub codebase. The vulnerabilities were exploited in July, but the findings are only being disclosed publicly now. OpenAI has since fixed the vulnerabilities and paid the researchers $6,500 through its bug bounty program.