Keynote: Guardians of the AI Era: Navigating the Cybersecurity Landscape of Tomorrow
summary
Argues that AI shifts enterprise threat models from deterministic features to non‑deterministic autonomous agents that can fetch internet data, act on it and be manipulated; demos and papers cited show image/audio instruction‑injection and alignment bypasses, and an AutoGPT example installs Node from StackOverflow to run code. Warns that model and vector DB poisoning enables exfiltration. Announces DARPA AIC: ~$20M total, five semi‑finalists get $2M, $4M first prize, Defcon 2024/2025 timeline with Anthropic Google Microsoft and OpenAI tooling available.
AI-written summary · about
chapters (24)
- 0:00Conference opening
- 1:57First-time attendees
- 4:05Global cyber threats
- 5:52AI economic impact
- 8:41AI regulations shift
- 11:34Keynote speaker Aaria
- 13:24AI evolution to general
- 16:11Corporate AI arms race
- 20:14Tech shift security lessons
- 23:27AI threat model changes
- 26:19ChatGPT adoption skepticism
- 29:54Autonomous AI business use
- 32:06Multimodal AI limits
- 35:40AI identity management
- 37:24AI model data risks
- 40:53AI agent exploitation vectors
- 42:16AI model alignment risks
- 44:00AI impact on security pros
- 45:24AI forensic analysis
- 47:30AI security key takeaways
- 49:21AI research collaboration gaps
- 50:46DARPA AI Cyber Challenge
- 51:49DARPA AI Cyber Challenge goals
- 59:54Closing remarks
related talks
19mKeynote: Towards Building Safe and Secure Agentic AI
01h 24mOffensive Cybersecurity by AI Promises and Pitfalls
38mKeynote: Thriving in the Age of AI
13mKeynote: Agentic AI Is Running Your Infrastructure
49mThe Intersection of AI and Cybersecurity
37mFundamentals of AI safety and security
16mKeynote: Navigating Turbulent Waters of AI: Building a Secure and Safe Foundation
53mHow AI is impacting cybersecurity. Does it?