Sponsored Demo: The True Cost of Building at Machine Speed
summary
Frontier models are generating exploits at machine speed, producing 10-50x more code and finding roughly 20,000 high‑critical vulnerabilities with only ~6% disclosed so far. CI/CD pipelines face mean time to exploit drops to minutes or hours, enabling AI-powered malware and chained exploits that legacy scanners miss. Proposed responses include a trusted clearinghouse for coordinated disclosure, enterprise-run forks of abandoned projects, and automated pre‑disclosure patching and non‑patch mitigations to push mean time to remediation negative.
AI-written summary · about
chapters (14)
- 0:00Open source risks with AI
- 1:22AI discovering vulnerabilities
- 3:07AI-driven development phases
- 5:12AI code volume and security
- 7:17Mythos effect on vulnerabilities
- 8:19Machine-speed exploits and defense
- 9:00Vulnerability disclosure limits
- 10:24Maintainer inbox noise
- 11:48Open source security futures
- 12:50Forking critical projects
- 13:52Vulnerability disclosure plans
- 14:55Patches before disclosure
- 16:41AI security paradox
- 17:26Q&A
related talks
39mThe AI Arms Race in Vulnerability Management, Who’s Winning?
27mHow to implement Security as Code?
37mPlatform and Security Engineering
26mWhen Models Write Malware: Defending Against On-demand Exploits
40mEmbracing Security - A Changing DevOps Landscape
45mExemplars, Laggards, & The Cautious: A Data-driven Look At High-velocity Software...
20mEvolving from patch management to risk mitigation
29mThe vulnerability apocalypse: AI broke the security playbook