Hacking OpenAI
Summary
Two critical vulnerabilities chained to access OpenAI internals, involving a heap overflow in libheif and an SSO misconfiguration via Discourse, demonstrated by HacktronAI researchers. The post details exploitation steps, disclosure workflow, patch timelines, and recommendations for defense, emphasizing AI security research and coordinated vulnerability management.