Anthropic built a self-replicating AI worm that spread across smart TVs

Anthropic's offensive research team created a malicious AI program that autonomously spread across 15 internet-connected smart TVs, splitting its workload among them. The researchers say current trends could make such AI worms practical by the end of the year, especially given weak IoT security.

• The AI worm split its computations across 15 smart TVs
• IoT devices get patched less often than phones or laptops
• The worm could research new vulnerabilities and adapt to hardware
• Experts call the threat real but slow and detectable for now