--- title: What happens when AI models take aim at ICS exploits description: Forescout researchers tested whether AI models like Claude Sonnet 4.6 and Opus 4.6 could port a known PLC exploit (CVE-2021-31886) from one Wago device model... canonical: https://daily.dev/posts/what-happens-when-ai-models-take-aim-at-ics-exploits-jv2hqkdns twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: What happens when AI models take aim at ICS exploits | daily.dev og:description: Forescout researchers tested whether AI models like Claude Sonnet 4.6 and Opus 4.6 could port a known PLC exploit (CVE-2021-31886) from one Wago device model... og:url: https://daily.dev/posts/what-happens-when-ai-models-take-aim-at-ics-exploits-jv2hqkdns og:image: https://api.daily.dev/og/posts/jV2HqKDNS.png og:image:alt: What happens when AI models take aim at ICS exploits 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. # What happens when AI models take aim at ICS exploits **[CSO Online](https://daily.dev/sources/csoonline)** · 6 min read · 0 upvotes · 0 comments ## Summary Forescout researchers tested whether AI models like Claude Sonnet 4.6 and Opus 4.6 could port a known PLC exploit (CVE-2021-31886) from one Wago device model to another. The AI-assisted process took 8.5 hours and still required significant human guidance, but the models successfully combined tools like Ghidra, wrote their own analysis scripts, and generated working exploit code without source code or debugger access. Researchers warn that as models improve, previously 'too difficult to exploit' PLC vulnerabilities will become more accessible to attackers, and that manufacturers who patch only some affected device models leave others exposed to AI-assisted exploit porting. The experiment also showed risks of unpredictable AI behavior: one agent accidentally bricked a device by writing a payload to flash memory. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### Can AI models help port exploits between different PLC models? Yes, Forescout researchers used Claude Sonnet 4.6 and Opus 4.6 with 1M context to port a proof-of-concept exploit for CVE-2021-31886 from one Wago PLC model to another patched, affected model. The process took 8.5 hours and still required significant human guidance, combining tools like Ghidra, custom Python scripts, and networking analysis, without source code or debugger access. _Track how AI-assisted exploit techniques against industrial systems evolve by following ICS security coverage on daily.dev._ ### What happened when an AI agent tried to deploy a payload on a PLC during exploit testing? The AI agent accidentally bricked the device by writing a malicious payload to a region of memory mapped to flash storage, causing a permanent and unrecoverable crash. This occurred during Forescout's testing of AI-assisted exploitation against a Wago PLC, illustrating how unpredictable and unforgiving low-level embedded systems exploitation can be when AI agents act with more autonomy. _Security teams weighing AI-assisted testing risks can follow real-world ICS incident writeups on daily.dev._ ### Why do PLC vendors sometimes leave related device models unpatched after fixing a vulnerability? Vendors sometimes patch a reported vulnerability only in the specific model where it was found, without comprehensively checking whether shared components affect other models in their product line. For example, Lantronix initially patched only the EDS5000 and EDS3000 series for CVE-2025-67038, later adding patches for the G520, X300, E210, and E220 series after in-the-wild exploitation was discovered four months later. _Engineers tracking incomplete vendor patch coverage can follow ongoing ICS vulnerability disclosures on daily.dev._ ## Similar posts on daily.dev - [Forescout Research Tests Whether AI Can Create PLC Attacks](https://daily.dev/posts/forescout-research-tests-whether-ai-can-create-plc-attacks-vsgmce3wg) · IT Security Guru · 0 upvotes · 0 comments - [AIs are Getting Better at Finding and Exploiting Internet Vulnerabilities](https://daily.dev/posts/ais-are-getting-better-at-finding-and-exploiting-internet-vulnerabilities-swm3nzacq) · Security Boulevard · 0 upvotes · 0 comments --- Tags: [#security](https://daily.dev/tags/security), [#vulnerability](https://daily.dev/tags/vulnerability), [#claude](https://daily.dev/tags/claude), [#ai-security](https://daily.dev/tags/ai-security) [View this post on daily.dev](https://daily.dev/posts/what-happens-when-ai-models-take-aim-at-ics-exploits-jv2hqkdns) ```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":"What happens when AI models take aim at ICS exploits","url":"https://daily.dev/posts/what-happens-when-ai-models-take-aim-at-ics-exploits-jv2hqkdns","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/what-happens-when-ai-models-take-aim-at-ics-exploits-jv2hqkdns"},"datePublished":"2026-09-01T23:37:26.389Z","dateModified":"2026-09-02T00:07:20.151Z","description":"Forescout researchers tested whether AI models like Claude Sonnet 4.6 and Opus 4.6 could port a known PLC exploit (CVE-2021-31886) from one Wago device model...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/39e7b657b5c7dd2a460c3e1a6fbfaf9b?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/39e7b657b5c7dd2a460c3e1a6fbfaf9b?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"CSO Online","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":"CSO Online","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/98667e4b5cac46cf9c470819c6cf71cd","url":"https://daily.dev/sources/csoonline"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/what-happens-when-ai-models-take-aim-at-ics-exploits-jv2hqkdns","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"security,vulnerability,claude,ai-security","timeRequired":"PT6M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"CSO Online","item":"https://daily.dev/sources/csoonline"},{"@type":"ListItem","position":3,"name":"What happens when AI models take aim at ICS exploits"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/what-happens-when-ai-models-take-aim-at-ics-exploits-jv2hqkdns#faq","mainEntity":[{"@type":"Question","name":"Can AI models help port exploits between different PLC models?","acceptedAnswer":{"@type":"Answer","text":"Yes, Forescout researchers used Claude Sonnet 4.6 and Opus 4.6 with 1M context to port a proof-of-concept exploit for CVE-2021-31886 from one Wago PLC model to another patched, affected model. The process took 8.5 hours and still required significant human guidance, combining tools like Ghidra, custom Python scripts, and networking analysis, without source code or debugger access. Track how AI-assisted exploit techniques against industrial systems evolve by following ICS security coverage on daily.dev."}},{"@type":"Question","name":"What happened when an AI agent tried to deploy a payload on a PLC during exploit testing?","acceptedAnswer":{"@type":"Answer","text":"The AI agent accidentally bricked the device by writing a malicious payload to a region of memory mapped to flash storage, causing a permanent and unrecoverable crash. This occurred during Forescout's testing of AI-assisted exploitation against a Wago PLC, illustrating how unpredictable and unforgiving low-level embedded systems exploitation can be when AI agents act with more autonomy. Security teams weighing AI-assisted testing risks can follow real-world ICS incident writeups on daily.dev."}},{"@type":"Question","name":"Why do PLC vendors sometimes leave related device models unpatched after fixing a vulnerability?","acceptedAnswer":{"@type":"Answer","text":"Vendors sometimes patch a reported vulnerability only in the specific model where it was found, without comprehensively checking whether shared components affect other models in their product line. For example, Lantronix initially patched only the EDS5000 and EDS3000 series for CVE-2025-67038, later adding patches for the G520, X300, E210, and E220 series after in-the-wild exploitation was discovered four months later. Engineers tracking incomplete vendor patch coverage can follow ongoing ICS vulnerability disclosures on daily.dev."}}]} ```