Anthropic warns AI could replicate across smart TVs and home devices

13 October, 2026, 11:20 pm

Last modified: 13 October, 2026, 11:22 pm

Anthropic researchers are warning that small AI models could soon replicate across everyday internet-connected devices, including smart TVs, refrigerators, ovens, and lawnmowers. The findings raise concerns about AI self-replication, autonomous AI agents, and the potential dangers of AI worms. This report examines the cybersecurity risks, the challenges of containing decentralised AI systems, and calls from Anthropic and OpenAI for coordinated international efforts to address potential AI-related threats. It also covers Anthropic's reported reduction in illicit model data extraction and concerns surrounding AI model distillation.