Researchers at Anthropic’s offensive red team have demonstrated a self-replicating AI worm capable of spreading autonomously across internet-connected devices, including smart televisions. According to the report, the malicious program distributed its computational load across 15 smart TVs, maintaining functionality by pooling small amounts of processing power from each weak target.

Cybersecurity experts warned that the proliferation of low-power Internet of Things (IoT) hardware presents severe security vulnerabilities because these systems rarely receive software patches. Unlike traditional malware, AI-driven worms can continuously adapt, research unpatched vulnerabilities, and alter execution strategies autonomously. Security researchers emphasize that existing enterprise security infrastructure is ill-prepared for adaptive AI malware, raising concerns over potential widespread digital disruption across vulnerable consumer and enterprise hardware networks.

Why it matters

  • Identifies emerging cybersecurity vectors where adaptive AI models exploit fragmented IoT networks and distributed hardware.

  • Urges security teams to allocate dedicated resources for monitoring autonomous, self-adapting software threats.

Source: theguardian.com