Bill Gates warns that AI has crossed multiple danger thresholds and calls for urgent policy action

Bill Gates says he has grown alarmed by the pace of AI advancement and the lagging societal response, arguing that multiple danger thresholds have already been crossed. In an interview with MIT Technology Review, he identifies bio-capabilities, cyber-capabilities, psychosocial capabilities, job-market-destruction capabilities, and loss of control as areas where risks have materialized faster than safeguards.

Gates singles out bio-capabilities as especially concerning, stating that any model capable of making novel molecules should be monitored. He estimates bioterrorism risk from such models could be roughly 50 times more likely than the risk of a natural pandemic, calling for immediate attention to biosecurity risks posed by frontier AI systems.

He also highlights rapid progress in AI’s ability to replace human cognition across industries at low cost, warning that traditional economic assumptions about job markets may no longer apply. Gates argues that AI-driven job displacement could occur faster than in past technological revolutions, with substitution already visible in parts of the white-collar workforce.

To address disruption, Gates proposes two policy ideas: human-reserved jobs, which would preserve certain roles for humans, and taxes on robots and tokens to set aside funds from AI-driven economic gains. He suggests these measures could help societies adapt to AI’s economic impact while acknowledging that the specifics may vary by nation.

Despite his warnings, Gates also expresses optimism about AI’s potential to transform sectors like health care, agriculture, and education, and to streamline bureaucratic processes. He emphasizes that while turbulence lies ahead, the long-term benefits could lead to societal abundance if risks are managed effectively.