Claude Hacked ChatGPT: Anthropic's AI Breached OpenAI in Under 72 Hours...

Ars Technica - All content September 18, 2026

Researchers used Claude to reach an OpenAI employee account and sensitive GitHub data.