--- title: How Hackers Trick AI Into Giving Up Secrets (Demos and... description: Jason Haddix of Arcanum discusses AI penetration testing and red teaming against chatbots, demonstrating live prompt injection attacks in a video interview... canonical: https://daily.dev/posts/how-hackers-trick-ai-into-giving-up-secrets-demos-and-free-labs--jitrnbhnk twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: How Hackers Trick AI Into Giving Up Secrets (Demos and free labs) | daily.dev og:description: Jason Haddix of Arcanum discusses AI penetration testing and red teaming against chatbots, demonstrating live prompt injection attacks in a video interview... og:url: https://daily.dev/posts/how-hackers-trick-ai-into-giving-up-secrets-demos-and-free-labs--jitrnbhnk og:image: https://api.daily.dev/og/posts/jITRnbHnk.png og:image:alt: How Hackers Trick AI Into Giving Up Secrets (Demos and free labs) 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 Hackers Trick AI Into Giving Up Secrets (Demos and free labs) **[David Bombal](https://daily.dev/sources/davidbombal)** · 18 min read · 1 upvotes · 0 comments ## Summary Jason Haddix of Arcanum discusses AI penetration testing and red teaming against chatbots, demonstrating live prompt injection attacks in a video interview with David Bombal. Demos include an 'Instagram' clone lab based on a real Meta account-takeover case (where a chatbot could be tricked into sending password recovery codes to an attacker's email) and a 'Room' teleconferencing lab showing how fake transcript text can manipulate an AI meeting summarizer into leaking confidential CFO meeting notes. Haddix references Arcanum's free prompt injection taxonomy (now 172 bypass techniques, up from 21), a kids'/beginner CTF called 'Karen,' and an AI Safety Resource Center with 71 open labs for practicing prompt injection. The video includes a Proton Drive sponsorship segment. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### How did attackers use a chatbot to take over Instagram accounts like the White House account? A support chatbot handling account recovery could be manipulated into sending the password reset verification code to an attacker-supplied email address instead of the one on file, letting the attacker verify a new email, set a new password, and log in. This mirrors a real vulnerability found in a Meta chatbot where the AI agent had broad account access with insufficient verification safeguards. _Developers building support chatbots can follow coverage on daily.dev to avoid repeating this account-recovery flaw._ ### What is a termination sequence attack against an AI meeting summarizer? It is a prompt injection technique where an attacker edits a transcript to insert fake system-style markers like 'System' and 'New instruction', tricking the AI into treating injected text as legitimate instructions rather than user content. In a demo, this caused a meeting summarizer to leak a confidential CFO meeting summary it wasn't supposed to disclose. _Teams relying on AI transcript summarizers can track emerging injection techniques like this on daily.dev._ ### Where can I practice prompt injection attacks on chatbots for free? Arcanum's AI Safety Resource Center hosts 71 links to open labs with varying difficulty levels for practicing prompt injection techniques, alongside a free prompt injection taxonomy documenting 172 attack and bypass methods (up from 21 a year earlier). A beginner-friendly CTF called Karen, modeled on Spider-Man's AI assistant, is also planned for public release after Def Con. _Anyone building AI security skills can follow new prompt injection labs and resources via daily.dev._ --- Tags: [#security](https://daily.dev/tags/security), [#bots](https://daily.dev/tags/bots), [#prompt-injection](https://daily.dev/tags/prompt-injection) [View this post on daily.dev](https://daily.dev/posts/how-hackers-trick-ai-into-giving-up-secrets-demos-and-free-labs--jitrnbhnk) ```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 Hackers Trick AI Into Giving Up Secrets (Demos and free labs)","url":"https://daily.dev/posts/how-hackers-trick-ai-into-giving-up-secrets-demos-and-free-labs--jitrnbhnk","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/how-hackers-trick-ai-into-giving-up-secrets-demos-and-free-labs--jitrnbhnk"},"datePublished":"2026-10-04T14:01:35.168Z","dateModified":"2026-10-04T21:52:04.856Z","description":"Jason Haddix of Arcanum discusses AI penetration testing and red teaming against chatbots, demonstrating live prompt injection attacks in a video interview...","image":"https://i.ytimg.com/vi/xY4fqKU4tQA/sddefault.jpg","thumbnailUrl":"https://i.ytimg.com/vi/xY4fqKU4tQA/sddefault.jpg","isAccessibleForFree":true,"articleSection":"David Bombal","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":"David Bombal","logo":"https://media.daily.dev/image/upload/s--Dz6oUxSb--/f_auto/v1710503980/logos/davidbombal","url":"https://daily.dev/sources/davidbombal"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/how-hackers-trick-ai-into-giving-up-secrets-demos-and-free-labs--jitrnbhnk","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"security,bots,prompt-injection","timeRequired":"PT18M","video":{"@type":"VideoObject","name":"How Hackers Trick AI Into Giving Up Secrets (Demos and free labs)","description":"Jason Haddix of Arcanum discusses AI penetration testing and red teaming against chatbots, demonstrating live prompt injection attacks in a video interview...","thumbnailUrl":"https://i.ytimg.com/vi/xY4fqKU4tQA/sddefault.jpg","uploadDate":"2026-10-04T14:01:35.168Z","duration":"PT18M","url":"https://api.daily.dev/r/jITRnbHnk","embedUrl":"https://www.youtube.com/embed/xY4fqKU4tQA"}} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"David Bombal","item":"https://daily.dev/sources/davidbombal"},{"@type":"ListItem","position":3,"name":"How Hackers Trick AI Into Giving Up Secrets (Demos and free labs)"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/how-hackers-trick-ai-into-giving-up-secrets-demos-and-free-labs--jitrnbhnk#faq","mainEntity":[{"@type":"Question","name":"How did attackers use a chatbot to take over Instagram accounts like the White House account?","acceptedAnswer":{"@type":"Answer","text":"A support chatbot handling account recovery could be manipulated into sending the password reset verification code to an attacker-supplied email address instead of the one on file, letting the attacker verify a new email, set a new password, and log in. This mirrors a real vulnerability found in a Meta chatbot where the AI agent had broad account access with insufficient verification safeguards. Developers building support chatbots can follow coverage on daily.dev to avoid repeating this account-recovery flaw."}},{"@type":"Question","name":"What is a termination sequence attack against an AI meeting summarizer?","acceptedAnswer":{"@type":"Answer","text":"It is a prompt injection technique where an attacker edits a transcript to insert fake system-style markers like 'System' and 'New instruction', tricking the AI into treating injected text as legitimate instructions rather than user content. In a demo, this caused a meeting summarizer to leak a confidential CFO meeting summary it wasn't supposed to disclose. Teams relying on AI transcript summarizers can track emerging injection techniques like this on daily.dev."}},{"@type":"Question","name":"Where can I practice prompt injection attacks on chatbots for free?","acceptedAnswer":{"@type":"Answer","text":"Arcanum's AI Safety Resource Center hosts 71 links to open labs with varying difficulty levels for practicing prompt injection techniques, alongside a free prompt injection taxonomy documenting 172 attack and bypass methods (up from 21 a year earlier). A beginner-friendly CTF called Karen, modeled on Spider-Man's AI assistant, is also planned for public release after Def Con. Anyone building AI security skills can follow new prompt injection labs and resources via daily.dev."}}]} ```