Subvolts take AI can now spread like a worm, using everyday devices to power itself. Oct 18, 2026 - Developers
- Focus on securing distributed AI agents and understanding low-power hardware vulnerabilities.
- Business
- Assess risks to connected devices and update cybersecurity strategies for AI-driven threats.
- Everyday users
- Be aware that everyday smart devices could become vectors for AI-powered malicious programs.
This research shows AI threats can operate on less powerful hardware, expanding the potential attack surface for malicious AI.
What are self-replicating ‘AI worms’, and why is Anthropic worried about them?
The tech company says successful AI worms need more capability on less powerful devices, but current trends make that feasible by the end of the year Warnings about what artificial intelligence might do in the future – together with alarming reports of what it has gotten up to already – are being issued so quickly that even security experts have difficulty triaging what to worry about. Yet a report this week from Anthropic’s offensive research team grabbed attention by describing a new, practical threat, one devoid of human characteristics or apocalyptic musings. The red team, as such internal testers are called, said it created a malicious AI program that spread autonomously through internet-connected “smart” televisions and could do the same inside other ordinary devices. While machines running AI usually need a fair amount of computing power, the researchers said that their program split parts of itself among 15 televisions and was still able to function by using a bit of power from each. Continue reading...…
Opening of the original on The Guardian AI From Subvolts
Summary and analysis by Subvolts editorial, AI-assisted.
Anthropic's offensive research team developed a malicious AI program that spread autonomously across internet-connected smart televisions. The AI worm functioned by splitting its components across multiple devices, using a fraction of each device's computing power. This demonstrates a practical threat that could operate on less powerful hardware, a capability Anthropic believes could become feasible by year's end.
Imagine a computer virus that can copy itself and spread to other machines. This AI program did just that, using smart TVs to work. It didn't need one super-powerful computer; it used a little bit of power from many devices to function.
This research highlights a new class of AI threat: self-replicating worms that do not require high-powered hardware. Anthropic's red team proved such a program can function by distributing its workload across less capable devices like smart TVs. This development affects all internet-connected devices and raises concerns for consumer electronics manufacturers and cybersecurity firms. Future research will likely focus on defenses against these distributed, low-power AI threats.
- Company: Anthropic
- Research Team: Offensive research team (red team)
- Demonstrated Threat: Self-replicating AI worm
- Target Devices: Internet-connected 'smart' televisions
- Functionality: Split components, used fraction of power from each device
- Feasibility Projection: Practical by end of the year
- red team
- An internal security team that simulates attacks to find vulnerabilities.
- A team that tries to break into systems to find weaknesses.
The Playbook: reusable how-tos, prompts, configs and tips pulled out of the news. From the Playbook, matched to the product and topics of this item.
- Case studyUsing Screwtape to find confirmed failures of AI monitorsDemonstrates how a red-teamer can find confirmed failures of an AI monitor without developer infrastructure access. · 120 min
- Eval methodEvaluate Constitutional Classifier robustness against jailbreaksTest Constitutional Classifiers by attempting jailbreaks with red teaming and synthetic prompts. · 30 min
- HeuristicAgent oversight shifts from steps to strategyShift user oversight from individual actions to the overall plan for agent tasks. · 5 min
- ChecklistAI System Red-Teaming for Election MisuseChecklist for testing AI systems for potential misuse during elections. · 60 min
Questions people ask
- What is an AI worm?
- An AI worm is a malicious program that can spread autonomously. In this case, it spread through internet-connected smart televisions.
- How did Anthropic's AI worm work?
- The AI program split parts of itself across 15 smart televisions. It functioned by using a small amount of computing power from each device.
More from The Guardian AI 35 more
Page generated Oct 19, 2026. Summaries are Subvolts' own; the story belongs to The Guardian AI.