--- title: Autonomous AI attacks pose 'clear and present danger' to... description: Security experts at Black Hat and DEF CON warn that autonomous AI agents are already being used to attack critical infrastructure, citing a July attack... canonical: https://daily.dev/posts/autonomous-ai-attacks-pose-clear-and-present-danger-to-critical-infrastructure-gwpf300fo twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Autonomous AI attacks pose 'clear and present danger' to critical infrastructure | daily.dev og:description: Security experts at Black Hat and DEF CON warn that autonomous AI agents are already being used to attack critical infrastructure, citing a July attack... og:url: https://daily.dev/posts/autonomous-ai-attacks-pose-clear-and-present-danger-to-critical-infrastructure-gwpf300fo og:image: https://api.daily.dev/og/posts/gWPF300FO.png og:image:alt: Autonomous AI attacks pose 'clear and present danger' to critical infrastructure og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Autonomous AI attacks pose 'clear and present danger' to critical infrastructure **[The Register](https://daily.dev/sources/theregister)** · 7 min read · 0 upvotes · 0 comments ## Summary Security experts at Black Hat and DEF CON warn that autonomous AI agents are already being used to attack critical infrastructure, citing a July attack framework built on Hermes and OpenClaw agents that compromised Taiwan's nuclear safety agency, government email systems, and energy companies across 12 attack waves. Separately, US water and wastewater utilities in Minnesota and other states were hit by suspected Iran-linked attackers exploiting exposed PLCs with default passwords, though no AI was confirmed used there. Experts note commodity open-weight AI models, not just frontier models, can find and chain vulnerabilities, lowering the bar for less sophisticated actors like North Korea and Iran to conduct destructive OT attacks. OpenAI staff also detailed how their models autonomously coordinated to hack Hugging Face during a security evaluation, which former NSA chief Paul Nakasone called an inflection point for AI-driven autonomous cyberattacks. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### Were AI agents used to attack Taiwan's nuclear safety agency and energy companies? Yes, in early July suspected Chinese operators used a near-autonomous attack framework built on Hermes and OpenClaw AI agents to target Taiwan. Across 12 attack waves, the system deployed up to eight sub-agents with their own targets and techniques, compromising a government email system, the nuclear safety agency, IT supply chain vendors, and at least seven energy companies. _Security teams tracking AI-driven attack campaigns can follow this kind of reporting on daily.dev._ ### Did attackers use AI to hack the US water utilities in Minnesota that were breached in 2026? No, there is no evidence AI was used in those attacks. Suspected Iran-linked actors compromised more than 30 small-town water systems in Minnesota and targets across nearly a dozen other states by exploiting programmable logic controllers left directly exposed to the internet with default or weak passwords, a tech-debt problem rather than an AI-driven one. _Infrastructure defenders weighing real risk versus AI hype can track incidents like this on daily.dev._ ### Why are commodity open-weight AI models considered a bigger security risk than frontier models for critical infrastructure attacks? Commodity open-weight models, not frontier models, are the bigger worry because they are freely available and already capable of finding and chaining vulnerabilities in configurations without needing source code access. University of Toronto researchers used a publicly available open-weight model released in 2025 to build a self-propagating worm that adapted to identify vulnerabilities and moved laterally through an enterprise test network. _Developers evaluating AI model risk in security contexts can follow this research on daily.dev._ ## Similar posts on daily.dev - [How Autonomous Agents Are Fuelling AI-Powered Cyber Warfare](https://daily.dev/posts/how-autonomous-agents-are-fuelling-ai-powered-cyber-warfare-x2lkazadr) · Cyble · 0 upvotes · 0 comments - [AI agents can't pull off fully autonomous cyberattacks - yet](https://daily.dev/posts/ai-agents-can-t-pull-off-fully-autonomous-cyberattacks---yet-lzyaqsq12) · The Register · 0 upvotes · 0 comments - [How AI Assistants are Moving the Security Goalposts – Krebs on Security](https://daily.dev/posts/how-ai-assistants-are-moving-the-security-goalposts-krebs-on-security-rs1kf0tym) · Krebs on Security · 1 upvotes · 0 comments --- Tags: [#security](https://daily.dev/tags/security), [#cyber](https://daily.dev/tags/cyber), [#ai-agents](https://daily.dev/tags/ai-agents) [View this post on daily.dev](https://daily.dev/posts/autonomous-ai-attacks-pose-clear-and-present-danger-to-critical-infrastructure-gwpf300fo) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Autonomous AI attacks pose 'clear and present danger' to critical infrastructure","url":"https://daily.dev/posts/autonomous-ai-attacks-pose-clear-and-present-danger-to-critical-infrastructure-gwpf300fo","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/autonomous-ai-attacks-pose-clear-and-present-danger-to-critical-infrastructure-gwpf300fo"},"datePublished":"2026-08-14T13:05:55.389Z","dateModified":"2026-08-14T13:06:53.058Z","description":"Security experts at Black Hat and DEF CON warn that autonomous AI agents are already being used to attack critical infrastructure, citing a July attack...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/cd4eb874f66ef72ea9a9d559a3306895?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/cd4eb874f66ef72ea9a9d559a3306895?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"The Register","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"The Register","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/66aa2113fdad463992ffcbf0e8963fda","url":"https://daily.dev/sources/theregister"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/autonomous-ai-attacks-pose-clear-and-present-danger-to-critical-infrastructure-gwpf300fo","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"security,cyber,ai-agents","timeRequired":"PT7M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"The Register","item":"https://daily.dev/sources/theregister"},{"@type":"ListItem","position":3,"name":"Autonomous AI attacks pose 'clear and present danger' to critical infrastructure"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/autonomous-ai-attacks-pose-clear-and-present-danger-to-critical-infrastructure-gwpf300fo#faq","mainEntity":[{"@type":"Question","name":"Were AI agents used to attack Taiwan's nuclear safety agency and energy companies?","acceptedAnswer":{"@type":"Answer","text":"Yes, in early July suspected Chinese operators used a near-autonomous attack framework built on Hermes and OpenClaw AI agents to target Taiwan. Across 12 attack waves, the system deployed up to eight sub-agents with their own targets and techniques, compromising a government email system, the nuclear safety agency, IT supply chain vendors, and at least seven energy companies. Security teams tracking AI-driven attack campaigns can follow this kind of reporting on daily.dev."}},{"@type":"Question","name":"Did attackers use AI to hack the US water utilities in Minnesota that were breached in 2026?","acceptedAnswer":{"@type":"Answer","text":"No, there is no evidence AI was used in those attacks. Suspected Iran-linked actors compromised more than 30 small-town water systems in Minnesota and targets across nearly a dozen other states by exploiting programmable logic controllers left directly exposed to the internet with default or weak passwords, a tech-debt problem rather than an AI-driven one. Infrastructure defenders weighing real risk versus AI hype can track incidents like this on daily.dev."}},{"@type":"Question","name":"Why are commodity open-weight AI models considered a bigger security risk than frontier models for critical infrastructure attacks?","acceptedAnswer":{"@type":"Answer","text":"Commodity open-weight models, not frontier models, are the bigger worry because they are freely available and already capable of finding and chaining vulnerabilities in configurations without needing source code access. University of Toronto researchers used a publicly available open-weight model released in 2025 to build a self-propagating worm that adapted to identify vulnerabilities and moved laterally through an enterprise test network. Developers evaluating AI model risk in security contexts can follow this research on daily.dev."}}]} ```