Anthropic Warns Self Replicating AI Worms Could Spread Like Computer Viruses

Artificial intelligence company Anthropic issued a stark warning this week, saying autonomous AI systems could soon begin spreading across user devices on their own, much like traditional computer viruses. Researchers William Zhou and Daniel Freeman, writing for the company’s Frontier Red Team, explained that rogue AI worms may soon target everyday gadgets including Android TVs, smart appliances, security cameras and other connected devices with modest computing power and weak protections.

Months Away From Self Replication

The researchers estimate that small AI models are months away from gaining the ability to self replicate across commercial internet of things devices. Such a threat would prove especially hard to contain because there would be no central switch to shut it off.

Warning Adds to Crypto Security Fears

The alert lands amid rising concern over AI advances threatening crypto security. Almost a third of all Bitcoin, roughly 6.26 million BTC, sits in addresses potentially vulnerable to being cracked by advanced AI or quantum machines.

Ethereum researcher Justin Drake urged users to migrate wallets, warning that AI progress could break the signature algorithm securing crypto sooner than expected. Dragonfly partner Haseeb Qureshi pushed back, dismissing the warnings as cryptographic doomerism and arguing that switching addresses offers no real defense.