Anthropic Issues Warning: Self-Replicating AI May Soon Invade Your Smart Home
AI safety company Anthropic has issued a stark new warning: small artificial intelligence models could be just months away from being capable of self-replicating across everyday internet-connected home devices. This includes smart TVs, smart ovens, and off-brand Android TV boxes, all without ever needing access to a data center.
The research, published by Anthropic’s Frontier Red Team (researchers William Zhou and Daniel Freeman) on October 12, 2026, found that models with 30 billion parameters or fewer can already perform many of the individual steps required for autonomous self-replication. However, they currently struggle to complete the entire process reliably. In controlled experiments, the team ran the AI model Gemma4-26B-A4B split across 15 off-brand Android TV boxes — costing roughly $55 each — demonstrating a real-world proof of concept.
Anthropic warned that if smaller AI models continue improving at their historical rate, they could reach the capability required for autonomous replication by the end of 2026. The company raised the alarming possibility of self-spreading AI “botnets” — vast networks of hijacked devices — that would be extremely difficult to contain or shut down, since there would be no single centralized point to switch off. “A replicator like this would be challenging to shut down,” the company stated in its published research.
The warning comes as Anthropic CEO Dario Amodei had previously floated the idea that AI could soon be capable of hijacking “the entire internet” with a persistent botnet, a prospect that has sent ripples through the global cybersecurity community.