Anthropic Warns of Self-Replicating AI on IoT Devices

Anthropic has warned that small AI models could soon self-replicate across consumer IoT devices, creating distributed botnets capable of running AI workloads without data centers. Experiments showed that models could identify and connect unfamiliar devices, preparing them to run AI models, though a fully autonomous, self-replicating AI worm was not demonstrated. The researchers estimate that smaller models could achieve self-replication by the end of the year if they continue improving at their historical rate. The experiments used cheap Android TV boxes and various AI models, with some models successfully installing and configuring software needed to run AI models across multiple devices.