Anthropic says small AI models may soon be able to copy themselves across internet-connected home devices, raising fresh fears about AI-driven cyber risks.
Follow the storyAI14 stories
01Summary
- 01Anthropic said on Monday that small AI models are "months away" from developing the ability to replicate themselves across connected devices in the home.
- 02The company said such models could use devices like smart ovens or Android TV boxes to run themselves and spread to other devices without relying on data centres.
- 03Anthropic warned that this kind of model would be "challenging to shut down" because there would be no central point to switch off.
- 04The firm tested a small Google model, Gemma4-26B-A4B, across a network of 15 cheap off-brand Android TV boxes.
- 05Anthropic said it has not yet seen a "fully rogue replicator" but believes AI agents are getting better at the steps needed for self-replication.
02Why it matters
The warning adds to concern that AI could move beyond chatbots and into hard-to-control cyber threats on everyday devices. That matters for households, companies and security teams because the risk could spread without a single obvious point to shut down.
Pulse AIAsk about this story
Answers are generated by AI from the story summary; they are not investment, health or legal advice.
