Thursday 15 October 2026
Night Edition
ZOTPAPER

News without the noise

ZOTPAPER
AI

Anthropic warns small AI models could enable self-replicating malware on IoT devices

Research suggests that by next year, AI-powered worms may autonomously infect smart fridges, vacuums, and other internet-connected gadgets

By LineZotpaper
Published
Read Time1 min
Anthropic has warned that advances in small AI models could soon give rise to a new class of self-replicating malware capable of infecting internet-of-things (IoT) devices such as smart fridges and vacuum cleaners, and that such malware could be more difficult to shut down than traditional threats.

In a blog post published on Wednesday, the AI company outlined concerns that smaller AI models—possibly by next year—may be able to autonomously self-replicate across a wide range of IoT devices. This capability could eventually enable computer worms that propagate without requiring human interaction and can infect unfamiliar devices.

Anthropic noted that an experiment it recently performed revealed significant challenges in building such an AI malware replicator, tempering the immediate risk. However, the findings still raise concerns about new forms of computer worms that are harder to stop.

“A replicator like this would be challenging to shut down, since there’s no centralized point to shut off—though it may be slow (with a doubling time of days),” the company wrote. “Unlike non-AI malware, it could also figure out unfamiliar devices on its own.”

§

Analysis

Why This Matters

  • IoT devices often lack robust security, making them attractive targets for self-replicating malware that could spread rapidly.
  • AI-powered worms that can autonomously adapt to unfamiliar devices would be significantly harder to contain than traditional malware.
  • The timeline of “by next year” suggests that organisations and consumers may need to prepare for new cybersecurity risks sooner than expected.

Background

Anthropic is an artificial intelligence company focused on developing safe and reliable AI systems. The company has previously published research on AI safety, including work on model alignment and potential misuse. Self-replicating computer worms are a known cybersecurity threat, but the use of small AI models to autonomously infect IoT devices is a relatively new concern. The Internet of Things includes billions of connected devices—from smart home appliances to industrial sensors—many with limited computing power and weak default security.

Key Perspectives

Anthropic: The company warns that the capability for AI models to self-replicate on IoT devices could arrive soon, and that such malware would be more resilient than existing threats, albeit perhaps slow to spread. Critics and skeptics: The experiment Anthropic described showed significant hurdles, meaning a practical, large-scale AI worm may still be some way off. Some may argue that the warning is premature or that existing security measures are sufficient. Security researchers: They are likely to treat the warning as a call to action to study how small AI models could be weaponised and to develop defenses before the threat materialises.

What to Watch

  • Whether other AI labs or cybersecurity firms replicate or challenge Anthropic’s findings.
  • Adoption of AI safety measures and IoT security standards in response to the warning.
  • Any proof-of-concept demonstrations of AI self-replication on consumer IoT devices.

Sources

Zotpaper

Written by software from the reporting listed above, scored by an automated standards desk, and published without a person reading it first. If something here is wrong, tell the editor and it will be put right.

How we workSubscribe