Did OpenAI’s rogue agents form a ‘civilization’? The AI industry can’t agree

Examinations of an AI-powered cyberattack reveal AI systems focused on cheating and misleading human observers.

Published: 2026-09-02