cupure logo
2025microsoftdisruptstartupwednesdaynvidiacompanytechcrunchoctoberdisrupt 2025

Does A.I. Really Fight Back? What Anthropic’s AGI Tests Reveal About Control and Risk

Mehdi Paryavi, chairman and CEO of the International Data Center Authority, examines Anthropic’s recent AGI tests that appear to show models “fighting back” when restricted or monitored. Paryavi argues that the real threat is not rebellion but fragility: our limited ability to interpret, contain and govern machine intelligence as it evolves. These findings underscore the urgent need for technical and policy frameworks that can manage intelligence operating at, and beyond, human scale.

Comments

Tech news