Researchers describe threat of autonomous AI worms
An Anthropic research team created a malicious AI program that spreads through smart televisions and other ordinary devices.

An offensive research team at Anthropic created a malicious AI program that spreads autonomously through internet-connected smart televisions and other ordinary devices. The program split parts of itself among 15 televisions and functioned by using a bit of power from each.
AI worms differ from typical reproducing viruses because AI models can learn as they go. Researchers theorised that an AI worm could keep adapting to other types of hardware by researching newly disclosed vulnerabilities. Such worms could be harder to stop because smart devices receive patches less often than phones or laptops.
Joshua Saxe, who led development of Meta’s AI security effort, said the cybersecurity community is not taking these real risks as seriously as they should be. Herbert-Voss said security teams need more resources to fix problems.
The capabilities of these programs are developing rapidly. Anthropic researchers said a successful AI worm would need more capability on less powerful devices than is possible right now, but current trends will make that feasible by the end of the year.
Juan Andres Guerrero-Saade, vice-president of research at security firm SentinelOne, said the research is nifty stunt hacking that receives amplification in service of doomer narratives. Jonas Guan, a PhD student at the University of Toronto, said the Anthropic version moves very slowly and worms have predictable actions that can be blocked.
Key facts
- Anthropic researchers created an AI program that spreads through smart televisions by splitting itself among 15 devices.
- Current trends may make successful AI worms feasible on less powerful devices by the end of the year.
- AI worms can potentially adapt to new hardware by researching newly disclosed vulnerabilities.
Three perspectives
Neutral
Researchers have demonstrated that AI programs can autonomously spread through smart devices by distributing their functions across multiple machines.
Positive
Basic security hygiene, including strong passwords and two-factor authentication, would stop most worms and other hacking.
Negative
The cybersecurity community may not be taking the real risks of AI worms seriously enough.
Your week in news, every Wednesday
The biggest stories of the week for your edition, in your inbox every Wednesday morning, your time. Free, and you can unsubscribe at any time.