Physix Frontier · News Briefing Card (Guardian Tech · Oct 18, 2026)
Anthropic: AI Worm Can Spread Autonomously Across Smart TVs
KEY FACTS
- Anthropic's red team released a report this week saying it created a malicious AI program.
- The program can spread autonomously through internet-connected smart TVs and run on other ordinary devices.
- The researchers split the program across 15 TVs, with each TV using only a small amount of compute to run it.
- The report notes that IoT devices such as smart TVs have weak security and are difficult to patch.
- Anthropic predicts that such AI worms will become feasible by the end of this year.
KEY DATA
15Number of infected TVs
PHYSIX OBSERVATION
Anthropic's demo, though criticized as a stunt, hits the long-standing soft spot of IoT security: devices like TVs update slowly and receive few patches, so once exploited by a worm, remediation costs are extremely high. This reminds the industry that AI security cannot focus only on large models; edge devices also need basic protection.
Source: Guardian Tech report
Physix Frontier