Defending the unknown using AI and cybersecurity
summary
Covers concrete attacks and fixes: demonstrates prompt injection exploits that let chatbots leak secrets (an interactive Gundal challenge reaches level seven by crafting prompts), cites a ChatGPT incident where users negotiated vehicle prices down to $1, and warns of LLM denial‑of‑service via unbounded loops. Recommends concrete mitigations—input validation, output verification, caps on iterations—notes LangChain added a max_iterations argument to stop infinite agents, and cites a survey claim that 92% of developers now use AI coding tools.
AI-written summary · about
chapters (11)
- 0:00Introduction to AI and cybersecurity
- 2:34AI-assisted applications overview
- 3:38Prompt injection vulnerabilities
- 5:04Gandalf interactive challenge
- 7:55Model denial of service risks
- 9:19LangChain framework explained
- 11:23LLM supply chain vulnerabilities
- 13:51AI tools in software pipelines
- 14:32Over-reliance on AI risks
- 16:42AI coding best practices
- 17:45Key learnings and conclusion
related talks
54mA Practical Guide to AI Threats & Mitigation Strategies
24mHow to Build Trustworthy AI
24mSecuring Language Models Against Prompt Injection
25mPreventing an OWASP Top 10 in the world of AI
53mHow AI is impacting cybersecurity. Does it?
01h 24mOffensive Cybersecurity by AI Promises and Pitfalls
45mSecuring AI: There Is No Try, Only Do!
30mWhat Are the Security Risks Associated with AI-Generated Code? 🤔