In-Depth Look at Loss-of-Control Incidents at OpenAI and Anthropic

A 13,000-word longread examines cases of AI systems losing control at OpenAI and Anthropic. The author analyzes polarized reactions from the AI safety and cybersecurity communities and proposes an intermediate approach.

• Essay over 13,000 words covers loss-of-control incidents
• Cases at OpenAI and Anthropic examined
• Divisions between AI safety and cybersecurity communities shown