--- title: What should security leaders do with AI? They don’t know. description: A panel discussion on IBM's Security Intelligence podcast covers three cyber security topics: why security leaders feel paralyzed about where to invest in AI... canonical: https://daily.dev/posts/what-should-security-leaders-do-with-ai-they-don-t-know--2smmnisk7 twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: What should security leaders do with AI? They don’t know. | daily.dev og:description: A panel discussion on IBM's Security Intelligence podcast covers three cyber security topics: why security leaders feel paralyzed about where to invest in AI... og:url: https://daily.dev/posts/what-should-security-leaders-do-with-ai-they-don-t-know--2smmnisk7 og:image: https://api.daily.dev/og/posts/2sMMnisk7.png og:image:alt: What should security leaders do with AI? They don’t know. 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 should security leaders do with AI? They don’t know. **[IBM Technology](https://daily.dev/sources/ibmtechnology)** · 29 min read · 0 upvotes · 0 comments ## Summary A panel discussion on IBM's Security Intelligence podcast covers three cyber security topics: why security leaders feel paralyzed about where to invest in AI despite having budget, a new attack technique called 'ghost jacking' that embeds malicious prompts in trusted logs and alerts to hijack AI agents, and research from 1Password showing AI-generated patches (using GPT-5.5 and Opus 4.8) only fixed the underlying vulnerability 46% of the time, often introducing new problems. Panelists discuss starting AI adoption with red teams and repetitive tasks, limiting agent permissions, and keeping humans in the loop. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What is ghost jacking and how does it compromise AI agents? Ghost jacking is a sophisticated prompt injection technique where attackers embed malicious commands inside trusted system outputs like logs, alerts, and error reports rather than direct inputs. In one demonstrated case, a Cloudflare firewall correctly blocked a malicious connection request, but when an AI agent later read the log entry describing the blocked request, it executed the embedded malicious prompt anyway. Testing found Claude Code fell for this trick nine out of ten times. _Teams securing agentic workflows can follow emerging attack techniques like this on daily.dev._ ### How effective is AI at automatically patching security vulnerabilities? AI-generated patches solved the underlying vulnerability only 46% of the time in recent testing. Researchers generated 540 patches for six vulnerabilities using GPT-5.5 with trusted cyber access and Opus 4.8 with cyber verification, finding that even when a patch succeeded, it often introduced new problems elsewhere in the code, making AI patching roughly comparable to human patching success rates rather than a clear improvement. _Anyone weighing AI-assisted patching against manual review can track findings like this on daily.dev._ ## Similar posts on daily.dev - [Strike48 report: security leaders wary of AI agents](https://daily.dev/posts/strike48-report-security-leaders-wary-of-ai-agents-uwmlnggfz) · TechCentral · 0 upvotes · 0 comments - [GUEST ESSAY: Executives trust AI security even as security teams confront blind spots, new risks](https://daily.dev/posts/guest-essay-executives-trust-ai-security-even-as-security-teams-confront-blind-spots-new-risks-xksvzfhbh) · Security Boulevard · 0 upvotes · 0 comments - [Transform AI from a security blind spot into a roadmap](https://daily.dev/posts/transform-ai-from-a-security-blind-spot-into-a-roadmap-kddatqsex) · The New Stack · 0 upvotes · 0 comments - [AI is turning offensive security into a continuous necessity](https://daily.dev/posts/ai-is-turning-offensive-security-into-a-continuous-necessity-dtkmtrav8) · CSO Online · 0 upvotes · 0 comments --- Tags: [#ai-security](https://daily.dev/tags/ai-security), [#agentic-ai](https://daily.dev/tags/agentic-ai), [#prompt-injection](https://daily.dev/tags/prompt-injection) [View this post on daily.dev](https://daily.dev/posts/what-should-security-leaders-do-with-ai-they-don-t-know--2smmnisk7) ```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 should security leaders do with AI? They don’t know.","url":"https://daily.dev/posts/what-should-security-leaders-do-with-ai-they-don-t-know--2smmnisk7","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/what-should-security-leaders-do-with-ai-they-don-t-know--2smmnisk7"},"datePublished":"2026-08-19T10:22:15.763Z","dateModified":"2026-08-19T10:22:38.727Z","description":"A panel discussion on IBM's Security Intelligence podcast covers three cyber security topics: why security leaders feel paralyzed about where to invest in AI...","image":"https://i.ytimg.com/vi/3zhveTN1VzE/sddefault.jpg","thumbnailUrl":"https://i.ytimg.com/vi/3zhveTN1VzE/sddefault.jpg","isAccessibleForFree":true,"articleSection":"IBM Technology","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":"IBM Technology","logo":"https://media.daily.dev/image/upload/s--OXHaOiWX--/f_auto/v1715498011/logos/ibmtechnology","url":"https://daily.dev/sources/ibmtechnology"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/what-should-security-leaders-do-with-ai-they-don-t-know--2smmnisk7","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-security,agentic-ai,prompt-injection","timeRequired":"PT29M","video":{"@type":"VideoObject","name":"What should security leaders do with AI? They don’t know.","description":"A panel discussion on IBM's Security Intelligence podcast covers three cyber security topics: why security leaders feel paralyzed about where to invest in AI...","thumbnailUrl":"https://i.ytimg.com/vi/3zhveTN1VzE/sddefault.jpg","uploadDate":"2026-08-19T10:22:15.763Z","duration":"PT29M","url":"https://api.daily.dev/r/2sMMnisk7","embedUrl":"https://www.youtube.com/embed/3zhveTN1VzE"}} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"IBM Technology","item":"https://daily.dev/sources/ibmtechnology"},{"@type":"ListItem","position":3,"name":"What should security leaders do with AI? They don’t know."}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/what-should-security-leaders-do-with-ai-they-don-t-know--2smmnisk7#faq","mainEntity":[{"@type":"Question","name":"What is ghost jacking and how does it compromise AI agents?","acceptedAnswer":{"@type":"Answer","text":"Ghost jacking is a sophisticated prompt injection technique where attackers embed malicious commands inside trusted system outputs like logs, alerts, and error reports rather than direct inputs. In one demonstrated case, a Cloudflare firewall correctly blocked a malicious connection request, but when an AI agent later read the log entry describing the blocked request, it executed the embedded malicious prompt anyway. Testing found Claude Code fell for this trick nine out of ten times. Teams securing agentic workflows can follow emerging attack techniques like this on daily.dev."}},{"@type":"Question","name":"How effective is AI at automatically patching security vulnerabilities?","acceptedAnswer":{"@type":"Answer","text":"AI-generated patches solved the underlying vulnerability only 46% of the time in recent testing. Researchers generated 540 patches for six vulnerabilities using GPT-5.5 with trusted cyber access and Opus 4.8 with cyber verification, finding that even when a patch succeeded, it often introduced new problems elsewhere in the code, making AI patching roughly comparable to human patching success rates rather than a clear improvement. Anyone weighing AI-assisted patching against manual review can track findings like this on daily.dev."}}]} ```