Empower AI with Security by Design
summary
Palo Alto Networks reports telemetry from 100,000+ customers and claims AI adoption hits 1 billion users in two years while observing an 890% increase in GenAI traffic since 2024. Examples include confidential IP uploaded to public GenAI, unauthorized plugins leaking access, and novel fast, scalable attack vectors. The recommended response is an integrated Prisma platform approach for inventorying AI assets, enforcing sanctioned-app policies, preventing data exfiltration, and closing blind spots left by siloed point solutions.
AI-written summary · about
chapters (8)
- 0:00Introduction to AI security
- 1:35AI adoption risks and threats
- 4:06AI benefits and productivity gains
- 6:52Security guardrails for AI adoption
- 9:02Points of compromise in AI systems
- 10:49Challenges with point solutions
- 12:58Protecting data in AI workflows
- 13:19Call to secure AI responsibly
related talks
37mReinforce AI security: Protecting AI applications, models, and data
28mSecuring AI by design
30mGenAI Opportunities and Challenges: Where 370 Enterprises Are Focusing Now
49mGenAI and Cybersecurity: Safeguarding the Digital Age
49mThe Intersection of AI and Cybersecurity
52mAI Governance: The Security Perspective
34m“You Can’t Just Turn It On”: The Hidden Work Behind Enterprise AI
42mSecuring the Invisible