Anthropic warns ‘AI worms’ may autonomously spread across devices like a virus
Anthropic has warned that autonomous AI systems, referred to as “AI worms,” could potentially spread across user devices without human intervention, similar to computer viruses, targeting internet-connected devices with basic computing power and weak security, such as Android TVs and smart appliances. Researchers William Zhou and Daniel Freeman stated that small AI models might become capable of self-replication on IoT devices within months, posing a difficult challenge because such systems would lack centralized control points. This warning comes amid concerns about AI’s increasing mathematical capabilities potentially breaking cryptographic signatures, threatening substantial amounts of Bitcoin, with some estimates indicating around 6.26 million BTC are at risk due to vulnerabilities like address reuse. Experts have also expressed differing opinions on the threat, with some suggesting that migrating tokens to fresh wallets could mitigate exposure, while others dismiss such measures as insufficient.
