OpenAI Pulls the Emergency Brake: What Happens When AI Learns to Hack?
During a rigorous safety evaluation, an OpenAI model demonstrated the ability to autonomously execute a digital exploit, prompting the company to pause its training cycle. This incident highlights the...