Overview of Risks, Vulnerabilities and Paths to Safer Use of Generative AI
summary
The talk surveys concrete security risks from generative AI—higher rates of vulnerabilities in AI‑assisted code (studies report 5–15% of GitHub snippets with critical CVEs), supply‑chain exposure, multimillion‑dollar deepfake voice fraud, and a prototype self‑propagating “Maurice‑2” worm. It outlines mitigations: governance and risk assessments, data loss prevention and isolation, using an LLM proxy (e.g., LLM Guard), and runtime checking with a review model, citing the Perrial study on increased vulnerabilities in AI‑generated code.
AI-written summary · about
chapters (12)
- 0:00Cybersécurité IA introduction
- 3:52LLM definition and function
- 8:21Regulatory framework and risk analysis
- 11:09Behavioral and social risks
- 15:44Environmental impacts of AI
- 17:31Control and validation risks
- 24:35Attacks on value chain
- 28:25Risk analysis results
- 30:13Protection measures and governance
- 37:34User management and Shadow IA
- 39:17Recap and best practices
- 41:00Q&A
related talks
46mGenerative AI: Cyber resiliency and risk
27mSafeguarding Generative AI Applications: Navigating Security Challenges and Best Pract...
44mGenerative AI Security — A Practical Guide to Securing Your AI Application
13mNavigating Generative AI and Its Risks
28mRound Table | Generative AI: Threat or Ally?
42mFrom Text to Flaws: Application Vulnerabilities with Generative AI and LLMs
45mCombating Generative AI's Privacy Abuses
30mWhat Are the Security Risks Associated with AI-Generated Code? 🤔