--- title: How Flawed Prompts Hamstring Powerful Reasoning Models description: Researchers from Zhejiang University and Alibaba have demonstrated a new vulnerability in AI reasoning models: deliberately crafted logically inconsistent... canonical: https://daily.dev/posts/how-flawed-prompts-hamstring-powerful-reasoning-models-tmhorgamd twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: How Flawed Prompts Hamstring Powerful Reasoning Models | daily.dev og:description: Researchers from Zhejiang University and Alibaba have demonstrated a new vulnerability in AI reasoning models: deliberately crafted logically inconsistent... og:url: https://daily.dev/posts/how-flawed-prompts-hamstring-powerful-reasoning-models-tmhorgamd og:image: https://api.daily.dev/og/posts/tmhoRgamd.png og:image:alt: How Flawed Prompts Hamstring Powerful Reasoning Models 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. # How Flawed Prompts Hamstring Powerful Reasoning Models **[IEEE Spectrum](https://daily.dev/sources/ieeespectrum)** · 5 min read · 0 upvotes · 0 comments ## Summary Researchers from Zhejiang University and Alibaba have demonstrated a new vulnerability in AI reasoning models: deliberately crafted logically inconsistent prompts can trigger excessive 'overthinking,' causing models to generate outputs up to 26 times longer than normal. Using an evolutionary genetic algorithm, the team corrupts the logical structure of math problems by swapping premises, deleting conditions, or mixing questions across problems. This black-box attack requires no access to model internals and works against major commercial models including DeepSeek-R1, Qwen3-Thinking, GPT-o3, and Gemini 2.5 Flash. At scale, this constitutes a denial-of-service vector that could degrade service for legitimate users. The researchers also showed malicious prompts transfer across models, increasing feasibility. The goal is to alert providers to this attack surface so defenses can be developed. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Similar posts on daily.dev - [20 approaches to writing better AI prompts](https://daily.dev/posts/20-approaches-to-writing-better-ai-prompts-tlrg0rbbe) · InfoWorld · 0 upvotes · 0 comments - [Schneier on Security](https://daily.dev/posts/schneier-on-security-s2fzr851g) · Schneier on Security · 0 upvotes · 0 comments - [The cost of thinking](https://daily.dev/posts/the-cost-of-thinking-zi8tusdea) · MIT News · 0 upvotes · 0 comments --- Tags: [#ai-security](https://daily.dev/tags/ai-security), [#cyber](https://daily.dev/tags/cyber), [#llm](https://daily.dev/tags/llm), [#prompt-engineering](https://daily.dev/tags/prompt-engineering) [View this post on daily.dev](https://daily.dev/posts/how-flawed-prompts-hamstring-powerful-reasoning-models-tmhorgamd) ```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":"How Flawed Prompts Hamstring Powerful Reasoning Models","url":"https://daily.dev/posts/how-flawed-prompts-hamstring-powerful-reasoning-models-tmhorgamd","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/how-flawed-prompts-hamstring-powerful-reasoning-models-tmhorgamd"},"datePublished":"2026-07-08T11:00:29.299Z","dateModified":"2026-10-15T02:21:44.631Z","description":"Researchers from Zhejiang University and Alibaba have demonstrated a new vulnerability in AI reasoning models: deliberately crafted logically inconsistent...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/ac7d4cbeb58f73c319b516538b2c334b?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/ac7d4cbeb58f73c319b516538b2c334b?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"IEEE Spectrum","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":"IEEE Spectrum","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/b123f5e8420e492ab36f6adb30c1793a","url":"https://daily.dev/sources/ieeespectrum"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/how-flawed-prompts-hamstring-powerful-reasoning-models-tmhorgamd","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-security,cyber,llm,prompt-engineering","timeRequired":"PT5M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"IEEE Spectrum","item":"https://daily.dev/sources/ieeespectrum"},{"@type":"ListItem","position":3,"name":"How Flawed Prompts Hamstring Powerful Reasoning Models"}]} ```