The AI bugpocalypse is here. Now what?
summary
Frontier models now find real vulnerabilities in production libraries at scale, not toys, with examples like Mythos and Fable and documented exploits such as Opus 46; model benchmarks show 20–40% introduction rate when writing code and industry data reports ~84% of developers using coding agents. Practical fixes are specific: rewrite critical libs to Rust, shift to memory-safe languages (preventing ~60–70% of unsafe bugs), deploy AI-enabled pre-PR scanning and visibility tooling, and harden open-source foundations before adversaries do.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 1:24AI coding tools risks
- 4:54Memory-safe languages
- 9:34Policy recommendations
- 13:27Autonomous AI agents
- 16:42Congressional testimony
- 19:15Call to action
related talks
29mThe vulnerability apocalypse: AI broke the security playbook
28mAll Your AI are Belong to Us
14mAI Sprawl: The Hidden Risk Threatening Your Enterprise Now
29mWhy did I break production? Is AI the sickness or the cure?
34mThe dual role of AI in Cybersecurity
37mHow to Break AI Systems (Before Someone Else Does)
39mThe AI Arms Race in Vulnerability Management, Who’s Winning?
26mSmarter Code, Sneakier Risks: Supply Chain Security in the Age of AI