What are self-replicating ‘AI worms’, and why is Anthropic worried about them?
Anthropic’s red team has demonstrated how a malicious AI program could spread across internet-connected devices, raising concerns about self-replicating “AI worms”. The risk matters because many smart devices have weak security and receive software updates less often than phones or laptops, making infections difficult to contain.
The researchers’ program divided its operations across 15 smart televisions, using a portion of each device’s computing power. They said a successful worm would need greater capability on less powerful hardware than is currently available, but current trends could make that feasible by the end of the year. Experts warn AI could help worms adapt to newly disclosed vulnerabilities; past worms such as WannaCry and NotPetya caused tens of billions of dollars in damage.
- Anthropic demonstrated an AI program operating across 15 smart TVs.
- Experts warn weakly secured devices could be hard to patch.
- More capable AI worms may become feasible by year-end.