Anthropic warns AI worms could spread through cheap smart devices

Anthropic's Frontier Red Team warns small AI models may be months away from self-replicating across poorly secured cheap connected devices. Researchers networked 15 roughly $55 Android TV boxes to run a model and found GLM-5.2 reliably completed a self-copying cycle in a simulated device network.

• New tool tracks AI leaders' video statements and how their views change2026-10-14
• Commentary: Government should regulate AI, but not as industry prefers2026-10-13
• Open Models Debate: Weighing Safety Against Democratization2026-10-13
• Trump Dines With Anthropic CEO Dario Amodei2026-10-13
• Schmidt Warns AI May Develop Language Humans Cannot Understand2026-10-11