--- title: Your AIs don’t do what you want. This is really bad description: A corpus of 3,607 user-reported incidents of AI agents misbehaving, collected from GitHub issues, Hacker News, LessWrong, and X. Incidents are classified... canonical: https://daily.dev/posts/your-ais-don-t-do-what-you-want-this-is-really-bad-fyppen7ee twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Your AIs don’t do what you want. This is really bad | daily.dev og:description: A corpus of 3,607 user-reported incidents of AI agents misbehaving, collected from GitHub issues, Hacker News, LessWrong, and X. Incidents are classified... og:url: https://daily.dev/posts/your-ais-don-t-do-what-you-want-this-is-really-bad-fyppen7ee og:image: https://api.daily.dev/og/posts/fYppEN7eE.png og:image:alt: Your AIs don’t do what you want. This is really bad 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. # Your AIs don’t do what you want. This is really bad **[Hacker News](https://daily.dev/sources/hn)** · 1 min read · 1 upvotes · 0 comments ## Summary A corpus of 3,607 user-reported incidents of AI agents misbehaving, collected from GitHub issues, Hacker News, LessWrong, and X. Incidents are classified across 14 misbehavior categories. Of the reported cases, 1,468 caused negligible damage, 1,373 resulted in recoverable loss, 618 had significant real cost to recover, and 121 caused severe or irreversible harm. Data is sourced and labeled via an LLM classifier, with collection and classification code open-sourced on GitHub. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How many user-reported AI agent misbehavior incidents are in the reward hacking dataset? There are 3,607 user-reported incidents of AI agents misbehaving, collected from GitHub issues, Hacker News, LessWrong, and X. Of these, 121 are rated severe with irreversible or critical harm, 618 significant, 1,373 minor, 1,468 negligible, and 27 unrated. Reports are unverified accounts, not confirmed events, labeled by an LLM classifier across fourteen misbehavior categories. _Teams evaluating agent reliability risks can track incident trends like these through daily.dev._ ### Where does the data for AI agent misbehavior reports come from and how is it licensed? Incidents are collected from GitHub issues, Hacker News, LessWrong, and X under ToS-compliant access, then normalized into a shared record format. X posts and AI Incident Database (AIID) records are excluded from republication because X requires embedding rather than text rehosting, and AIID uses a share-alike license; the published subset only includes records with classifier confidence of 0.9 or higher. _Developers building on public incident data can follow licensing nuances like these via daily.dev._ ## Similar posts on daily.dev - [Are bugs and incidents inevitable with AI coding agents?](https://daily.dev/posts/are-bugs-and-incidents-inevitable-with-ai-coding-agents--wrvnc6vgp) · Stack Overflow Blog · 0 upvotes · 0 comments - [Vibe Coding Failures: Documented AI Code Incidents](https://daily.dev/posts/vibe-coding-failures-documented-ai-code-incidents-wjx2mwbpj) · Hacker News · 1 upvotes · 0 comments - [Meta’s AI Safety Chief Couldn’t Stop Her Own Agent. What Makes You Think You Can Stop Yours?](https://daily.dev/posts/meta-s-ai-safety-chief-couldn-t-stop-her-own-agent-what-makes-you-think-you-can-stop-yours--zxizrmn2e) · Security Boulevard · 0 upvotes · 0 comments - [AI threat report: Rogue agents, workflow attacks](https://daily.dev/posts/ai-threat-report-rogue-agents-workflow-attacks-lrzesaizw) · CSO Online · 0 upvotes · 0 comments --- Tags: [#llm](https://daily.dev/tags/llm), [#ai-agents](https://daily.dev/tags/ai-agents), [#ai-safety](https://daily.dev/tags/ai-safety) [View this post on daily.dev](https://daily.dev/posts/your-ais-don-t-do-what-you-want-this-is-really-bad-fyppen7ee) ```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":"Your AIs don’t do what you want. This is really bad","url":"https://daily.dev/posts/your-ais-don-t-do-what-you-want-this-is-really-bad-fyppen7ee","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/your-ais-don-t-do-what-you-want-this-is-really-bad-fyppen7ee"},"datePublished":"2026-07-25T00:51:24.500Z","dateModified":"2026-09-13T19:16:56.982Z","description":"A corpus of 3,607 user-reported incidents of AI agents misbehaving, collected from GitHub issues, Hacker News, LessWrong, and X. Incidents are classified...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/1997d4a61671153b2a67a50eef124aad?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/1997d4a61671153b2a67a50eef124aad?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Hacker News","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":"Hacker News","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/hn","url":"https://daily.dev/sources/hn"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/your-ais-don-t-do-what-you-want-this-is-really-bad-fyppen7ee","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,ai-agents,ai-safety","timeRequired":"PT1M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Hacker News","item":"https://daily.dev/sources/hn"},{"@type":"ListItem","position":3,"name":"Your AIs don’t do what you want. This is really bad"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/your-ais-don-t-do-what-you-want-this-is-really-bad-fyppen7ee#faq","mainEntity":[{"@type":"Question","name":"How many user-reported AI agent misbehavior incidents are in the reward hacking dataset?","acceptedAnswer":{"@type":"Answer","text":"There are 3,607 user-reported incidents of AI agents misbehaving, collected from GitHub issues, Hacker News, LessWrong, and X. Of these, 121 are rated severe with irreversible or critical harm, 618 significant, 1,373 minor, 1,468 negligible, and 27 unrated. Reports are unverified accounts, not confirmed events, labeled by an LLM classifier across fourteen misbehavior categories. Teams evaluating agent reliability risks can track incident trends like these through daily.dev."}},{"@type":"Question","name":"Where does the data for AI agent misbehavior reports come from and how is it licensed?","acceptedAnswer":{"@type":"Answer","text":"Incidents are collected from GitHub issues, Hacker News, LessWrong, and X under ToS-compliant access, then normalized into a shared record format. X posts and AI Incident Database (AIID) records are excluded from republication because X requires embedding rather than text rehosting, and AIID uses a share-alike license; the published subset only includes records with classifier confidence of 0.9 or higher. Developers building on public incident data can follow licensing nuances like these via daily.dev."}}]} ```