Anthropic warns autonomous 'AI worms' could spread across vulnerable IoT devices

Anthropic's Frontier Red Team researchers William Zhou and Daniel Freeman warned that self-replicating autonomous AI systems, or 'AI worms,' could spread across everyday internet-connected devices like Android TVs and smart appliances without human intervention. They estimate small AI models may be months away from self-replicating on commercial IoT devices, posing a threat because there would be no centralized point to shut them off. This warning follows concerns that advanced AI models might break cryptographic signatures in crypto wallets, potentially threatening about one-third of the Bitcoin supply, which includes 6.26 million BTC in vulnerable addresses.