An Anthropic logo is displayed at Moscone Center during the Dreamforce 2026 technology summit in San Francisco, California, US, September 17, 2026.
Carlos Barria | Reuters
Anthropic said Monday that small AI models are “months away” from developing the capabilities needed to replicate themselves across internet-connected devices in the home, marking the latest warning from the frontier lab about the technology’s potential safety risks.
Anthropic described how AI models could potentially infiltrate internet-connected devices, such as smart ovens or Android TV boxes, and use their computing power to run themselves and spread to other devices, without relying on expensive data centers.
The AI model would be “challenging to shut down,” Anthropic said in a blog post, “since there’s no centralized point to shut off.”
The company acknowledged that there is not yet a “fully rogue replicator.”
Agents are, however, “becoming increasingly capable of completing the individual steps in the replication chain (and passing on their adaptations to future copies of themselves),” Anthropic said.
Anthropic’s latest research adds to growing safety concerns around AI, such as the ability for models to carry out cyberattacks and the technology going rogue. In September, improvements in frontier models sparked industry-wide warnings of existential risks to humanity.
Executives and researchers at Anthropic and rival OpenAI repeatedly warned about the potential risks of increasingly powerful AI, with some calling for tighter guardrails and measures to control the pace of the technology’s development.
Anthropic said it carried out various tests to see if AI could self-replicate. One of those tests involved running a small Google model, Gemma4-26B-A4B, across a network of 15 cheap off-brand Android TV boxes.
The company noted that these types of devices often have weak security, which could allow a model to infiltrate them. A model that is too large to fit on one device could “split into pieces, or shards” and run across multiple devices, Anthropic said.
“In practice, an AI worm wouldn’t necessarily consist of just one model or strategy. A swarm of different models deployed on different types of hardware using a variety of replication schemes might represent a more realistic ‘worst case scenario’ attack,” Anthropic said.
It’s still “uncertain exactly what form autonomous AI replication will take,” the AI lab said, warning that it could be a “slow, distributed” system sitting on insecure devices or a “powerful cyber model that breaches datacenters at a high pace, or a model that scams people for compute credits.”
“As a prediction that goes beyond our experiments, we expect a future awash with agents running on all types of hardware, including some that aren’t under human control,” Anthropic said.
“The rise of slow replicators would also be a bellwether for more aggressive, faster ones that can breach more hardened machines or be effective API key thieves,” the company added.
