Anthropic warns small AI models may soon self-replicate across connected devices
Anthropic warns small AI models could soon autonomously self-replicating across poorly secured smart devices, raising major containment and security concerns
AI firm Anthropic has warned that compact AI models could soon achieve autonomous self-replication across internet-connected devices without requiring centralized data centers.
Mythos and Claude artificial intelligence platform warns small AI models could soon autonomously self-replicating across poorly secured smart devices, raising major containment and security concerns.
Threat of AI self-replication:
In a newly published research evaluation, Anthropic found that models under 30 billion parameters can execute many of the isolated tasks required to spread across consumer hardware, such as smart TVs and appliances.
While complete end-to-end replication currently demands simplified or less restricted environments, historical performance trajectories suggest compact systems could master autonomous propagation capabilities.
Unlike traditional malware, self-replicating AI agents can independently analyze and adapt to unfamiliar network nodes.
Because these distributed systems operate independently of central cloud infrastructure, security experts emphasize they would lack a definitive kill switch.
Key experimental findings:
Anthropic has urged developers and manufacturers to significantly harden baseline security across consumer electronics to mitigate future loss-of-control risks.
In its experiments, Anthropic looked at whether AI models could find vulnerable devices, copy themselves into those devices, and continue spreading without human help. It found that small models under 30 billion parameters could complete the many steps needed for replication.
However, the firm found that such AI models also struggled to complete the entire sequence from start to finish.
In its simplified tests, where some security steps were removed, Anthropic found that GLM-5.2 and stronger models were capable of end-to-end replication.
Replicating models would be hard to shut down as there is no central switch-off point and, unlike non-AI malware, they could also figure out unfamiliar devices on their own, targeting data center processors, Anthropic warned.
Security and containment calls:
The firm focused on IoT devices like Android TVs, smart appliances, and other internet-connected gadgets in its research as these are typically poorly secured.
The leading American AI safety and research company setup also warned that if small models improved at their historical rate, they would be able to autonomously replicate by the end of this year, with small weight-based replicators following suit soon after.
Dario Amodei's firm Anthropic said on Monday its experiments suggested small AI models may be months away from being capable enough to self-replicate on commercial internet-of-things (IoT) devices, such as smart ovens and off-brand Android TV boxes.
Anthropic also said such autonomous AI systems were likely to replicate slowly and rely on widely available, relatively insecure computing resources.
The AI firm added that systems should be strengthened to make it harder for such AI agents to spread.
