Trust in the Age of AI Agents | Building Trustworthy AI
summary
Maps trust from the paranoia era to a confidence era and now a systemic reckoning, arguing that blind trust in autonomous agents creates real security gaps. Vulcan Cyber finds ChatGPT referencing non‑existent packages in ~100 of 400 queries; Trend Micro (July 2024) documents invented API methods; OpenClaw ships ~900 skills with ~20% flagged malicious and Maltbook breach exposes 1.5M API tokens and 35k emails. Modern CodeX auto‑generates and reruns tests until they pass. Code Registry analyzes 3 billion lines and sells VerifyYourCode ($9.99/mo or $25 one‑off) that checks security, licenses and tech debt without using AI.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 3:13Paranoia era of AI agents
- 10:55Confidence era and trust in AI
- 17:15Blind trust in autonomous agents
- 23:21Safer AI agent usage
- 28:41Summary
- 29:23Q&A
related talks
24mHow to Build Trustworthy AI
12mTrusting AI with Code: How Secure is AI-written code?
01h 24mOffensive Cybersecurity by AI Promises and Pitfalls
38mThe Trust Lifecycle for Safe, Secure and Reliable AI Agents
38mTrustworthy AI: The Key to Unlocking High-Stakes Applications
53mHow AI is impacting cybersecurity. Does it?
19mKeynote: Towards Building Safe and Secure Agentic AI
29mBuilding Trust in AI Agents