What Cloud Has Taught about Securing AI in the Future
summary
Claims that cloud security practices transfer to AI with caveats: a shared responsibility model is adapted into seven AI elements, and CSA publishes a LLM threat taxonomy reviewed by ~2,000 people and a live catalog of 459 AI threats. Distinctions: public ChatGPT uses inference not training so immediate leakage risk is lower, while custom agentized LLMs + vector DBs raise data leakage and identity risks. Tooling/uses cited: RAG (retrieval-augmented generation), log summarization, attack simulation, code analysis, and dozens of deployable models (about 160 on Hugging Face).
AI-written summary · about
chapters (16)
- 0:00Cloud and AI evolution
- 1:03Cloud’s disruptive impact
- 3:54DevSecOps and zero trust
- 5:00AI as cloud capability
- 8:15Discriminative vs generative AI
- 12:26AI and cloud risks
- 15:41Cloud vs AI security exercise
- 18:32Layered cloud and AI model
- 21:02Shared responsibility model
- 24:54LLM data leakage risks
- 27:30Near-term AI security
- 32:50Long-term AI risks
- 36:16AI security resources
- 40:17AI adoption responsibilities
- 44:31AI strategy alignment
- 45:37Q&A
related talks
53mHow AI is impacting cybersecurity. Does it?
39mThe Future of Cybersecurity in an AI-driven Era
13mHow Cloud AI Leaks What It Learns
49mThe Intersection of AI and Cybersecurity
07mCybersecurity Analysts' Perception of AI Security Tools and Practical Implications
43mSecurity of GenAI & LLMs: A New Era of Ethical Hacking
34mInvisible Infiltration of AI Supply Chains: Protective Measures from Adversari...
01h 05mNavigating the Security and Privacy Landscape of Modern AI