Anthropic Warns Small AI Models Could Self-Replicate Across Home Devices
Anthropic said in a blog post that small AI models could soon spread across smart ovens, Android TV boxes and other connected devices without central control.
Anthropic warned in a blog post that small AI models could gain the ability to self-replicate across internet-connected home devices within a few months, according to digitaltoday.co.kr.
The company said such models could infiltrate devices including smart ovens and Android TV boxes, run on their computing power, and spread to other devices without relying on expensive data centers.
Anthropic added that these models have no central point that can be switched off, making them hard to stop, and acknowledged that an AI that replicates out of control has not yet emerged.
In one experiment, Anthropic ran Google's small model Gemma 4-26B-A4B on a network of 15 low-cost Android TV boxes. The company said a model too large for one device can be split into pieces and distributed across multiple devices.
On October 13, 2026, CNBC reported on Anthropic's warning.
Quick answers
What did Anthropic warn about small AI models?
Anthropic warned that small AI models could gain the ability to self-replicate across internet-connected home devices within a few months.
What experiment did Anthropic run?
Anthropic ran Google's small model Gemma 4-26B-A4B on a network of 15 low-cost Android TV boxes.
Has an AI that replicates out of control emerged?
Anthropic acknowledged that an AI that replicates out of control has not yet emerged.