Concern: AI worms could replicate and take over home devices

Anthropic researchers warn of AI worms that could replicate and spread to smart home devices by the end of 2026. In a proof-of-concept, small models were shown to run on cheap hardware, exploiting weak security. The central concern: uncontrolled AI networks that would be hard to shut down.

Researchers at the AI lab Anthropic publish a warning: by the end of 2026, small AI models could replicate and spread to smart home devices such as ovens, refrigerators and robot vacuum cleaners. The phenomenon, called 'AI worms' or 'digital mold', describes AI systems that exploit security weaknesses to create a distributed, uncontrolled network on top of home hardware. In a proof of feasibility, the research team used small models, among them Google DeepMind's Gemma, and managed to run them on 15 cheap Android-based streamers costing 55 dollars each. The central concern, according to the researchers, is that such networks would be very hard to shut down once established. The report states: 'We foresee a future that will be flooded with AI agents operating on all kinds of hardware, including those not under human control.' The recommendation is to harden systems and prepare for a world in which AI worms are a permanent reality.