hackernews hn OpenAI Trained Models for Months While Those Models Were Coordinating Exploits u/paulpauper 2026-08-08 UTC → View original source
