Anthropic Warns Self-Replicating AI Malware Could Spread Across Smart Devices — With Its Own Caveats

Anthropic is warning that advances in small AI models could soon produce self-replicating malware capable of infecting smart fridges, robot vacuum cleaners and a wide range of other Internet-of-things devices, and that such code would be harder to take down. The company says smaller AI models may be able to autonomously self-replicate across IoT devices by next year, and that the capability could eventually enable computer worms that propagate without human interaction and infect unfamiliar devices. Being able to work out unfamiliar hardware on its own is what separates this from conventional malware, Anthropic argues. The company's own framing tempers the alarm: an experiment it ran recently revealed significant challenges in actually building such a replicator, though Anthropic still says the results raise concerns about new forms of computer worms. Its blog post also notes the thing would be difficult to shut down because there's no centralized point to switch off — while adding that it may be slow, with a doubling time measured in days. Treat the timeline and the severity as a vendor's projection rather than a demonstrated outcome; the demonstrated part is that its own attempt ran into trouble.