Behind the Curtain: The AI titans' biggest private fear

Axios reports that when AI leaders are asked in private what risk worries them most, the answer is strikingly consistent: an AI-engineered killer pathogen that spreads too quietly and too widely to stop. The concern is echoed in a new MIT FutureTech and University of Queensland study, in which two hundred seventy-two researchers ranked the top AI risks and put a twelve percent chance on AI's dangerous capabilities producing a catastrophic outcome by twenty thirty — defined as more than one million deaths or one hundred billion dollars in damage — with the odds climbing above twenty percent without mitigation efforts. Axios notes that just last month the heads of OpenAI, Anthropic, Google DeepMind, Microsoft, and Meta cosigned an open letter warning of AI-derived bioweapons and calling for stronger safeguards. The fear, as experts frame it, is that models trained on vast biological datasets could help a bad actor design a novel pathogen faster than detection and prevention systems can catch up.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton