--- title: Was I wrong about open models & cyber risks? | daily.dev description: Nathan Lambert revisits his earlier claim that open-weight AI models wouldn't cause a cyber disaster, after a wave of critical vulnerability disclosures dubbed... canonical: https://daily.dev/posts/was-i-wrong-about-open-models-cyber-risks--r00rvmv4l twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Was I wrong about open models & cyber risks? | daily.dev og:description: Nathan Lambert revisits his earlier claim that open-weight AI models wouldn't cause a cyber disaster, after a wave of critical vulnerability disclosures dubbed... og:url: https://daily.dev/posts/was-i-wrong-about-open-models-cyber-risks--r00rvmv4l og:image: https://api.daily.dev/og/posts/R00rVmv4L.png og:image:alt: Was I wrong about open models & cyber risks? 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. # Was I wrong about open models & cyber risks? **[Interconnects](https://daily.dev/sources/interconnects)** · 9 min read · 0 upvotes · 0 comments ## Summary Nathan Lambert revisits his earlier claim that open-weight AI models wouldn't cause a cyber disaster, after a wave of critical vulnerability disclosures dubbed 'Black October' and a major Binance hack. He still believes GLM-5.3, which he says crosses the capability threshold once feared for Claude Mythos, hasn't caused a step-change in harm a month after release. Epoch AI used GLM-5.3 to build a Firefox exploit that could exfiltrate files just by visiting a malicious site, showing real risk but requiring multiple steps to cause harm. Lambert now argues closed models have played a bigger role in recent hacks than open ones, and that open models remain too expensive to serve widely for attackers to rely on yet. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### Did GLM-5.3 actually cause a major increase in cyberattacks after its release as an open-weight model? There is little public evidence that much has changed in cyberattack severity more than a month after GLM-5.3's weight release, despite it reportedly crossing the capability threshold once feared for models like Claude Mythos. Epoch AI did use GLM-5.3 to build a proof-of-concept Firefox exploit that could exfiltrate files from a user's computer just by visiting a malicious website, showing real but multi-step risk rather than a sudden disaster. _Anyone weighing open-weight model risk can track emerging exploit research like this through daily.dev._ ### Are open-weight or closed AI models more responsible for recent cybersecurity incidents? Closed models have played a heavier role in recent hacks than open-weight models, based on available evidence, though confidence in this is low. Open models remain expensive to serve and compute for the largest ones is heavily sold out, making it more likely that any near-term open-model-driven attack would come through a poorly secured open model API rather than large-scale deployment. _Developers evaluating model risk for security-sensitive projects can follow this debate via daily.dev._ ## Similar posts on daily.dev - [The Cyber Risk Discourse is Broken](https://daily.dev/posts/the-cyber-risk-discourse-is-broken-w1ckoqv8l) · Interconnects · 0 upvotes · 0 comments - [Open-weight AI caught the frontier on capability. On safety, it didn’t](https://daily.dev/posts/open-weight-ai-caught-the-frontier-on-capability-on-safety-it-didn-t-onyvj8soa) · The Next Web · 0 upvotes · 0 comments - [Just say no to OpenAI and Anthropic](https://daily.dev/posts/just-say-no-to-openai-and-anthropic-ktuvq0lsn) · Hacker News · 5 upvotes · 1 comments --- Tags: [#security](https://daily.dev/tags/security), [#cyber](https://daily.dev/tags/cyber), [#open-source](https://daily.dev/tags/open-source), [#llm](https://daily.dev/tags/llm), [#ai-safety](https://daily.dev/tags/ai-safety) [View this post on daily.dev](https://daily.dev/posts/was-i-wrong-about-open-models-cyber-risks--r00rvmv4l) ```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":"Was I wrong about open models \u0026 cyber risks?","url":"https://daily.dev/posts/was-i-wrong-about-open-models-cyber-risks--r00rvmv4l","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/was-i-wrong-about-open-models-cyber-risks--r00rvmv4l"},"datePublished":"2026-10-30T17:42:26.785Z","dateModified":"2026-10-30T17:42:54.344Z","description":"Nathan Lambert revisits his earlier claim that open-weight AI models wouldn't cause a cyber disaster, after a wave of critical vulnerability disclosures dubbed...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/4d61c35e0edc0df19c1084fe44be31b6?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/4d61c35e0edc0df19c1084fe44be31b6?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Interconnects","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":"Interconnects","logo":"https://media.daily.dev/image/upload/s--BshHFZ2M--/f_auto,q_auto/v1780213203/logos/interconnects?_a=BAMAMiWQ0","url":"https://daily.dev/sources/interconnects"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/was-i-wrong-about-open-models-cyber-risks--r00rvmv4l","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"security,cyber,open-source,llm,ai-safety","timeRequired":"PT9M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Interconnects","item":"https://daily.dev/sources/interconnects"},{"@type":"ListItem","position":3,"name":"Was I wrong about open models \u0026 cyber risks?"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/was-i-wrong-about-open-models-cyber-risks--r00rvmv4l#faq","mainEntity":[{"@type":"Question","name":"Did GLM-5.3 actually cause a major increase in cyberattacks after its release as an open-weight model?","acceptedAnswer":{"@type":"Answer","text":"There is little public evidence that much has changed in cyberattack severity more than a month after GLM-5.3's weight release, despite it reportedly crossing the capability threshold once feared for models like Claude Mythos. Epoch AI did use GLM-5.3 to build a proof-of-concept Firefox exploit that could exfiltrate files from a user's computer just by visiting a malicious website, showing real but multi-step risk rather than a sudden disaster. Anyone weighing open-weight model risk can track emerging exploit research like this through daily.dev."}},{"@type":"Question","name":"Are open-weight or closed AI models more responsible for recent cybersecurity incidents?","acceptedAnswer":{"@type":"Answer","text":"Closed models have played a heavier role in recent hacks than open-weight models, based on available evidence, though confidence in this is low. Open models remain expensive to serve and compute for the largest ones is heavily sold out, making it more likely that any near-term open-model-driven attack would come through a poorly secured open model API rather than large-scale deployment. Developers evaluating model risk for security-sensitive projects can follow this debate via daily.dev."}}]} ```