The AI Arms Race in Vulnerability Management, Who’s Winning?
summary
AI cuts the median time from disclosure to exploitation to 4 hours median time-to-exploit and produces both noise and effective output: automated reports swamp triage while LLM agents achieve 87% success at <$9 on one dataset and Project Zero’s Big Sleep finds unknown SQLite bugs. Mitis/Mythis claims a ~70% exploit rate and Anthropic-style scans surface 500+ high-severity open-source flaws. Defenses shift to context engineering, remediation automation, supply-chain controls, and active deception to mislead agentic attackers.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:28AI arms race in vulnerability management
- 5:20AI-driven vulnerability milestones
- 11:00AI’s offensive and defensive edge
- 16:16Why attackers gain with AI
- 20:49AI impact on vulnerability reporting
- 23:16Defender strategies in AI era
- 27:09AI supply chain risks
- 30:39From vulnerabilities to root causes
- 31:42AI deception to outmaneuver attackers
- 34:10Summary and call to action
related talks
53mAI-equipped Threat Actors Versus AI-enhanced Cyber Tools: Who Wins?
29mThe vulnerability apocalypse: AI broke the security playbook
20mSecurity, and AI... so happy together?
20mCatch Attackers Before They Strike with AI-powered Threat Intelligence
13mThe Automation of Exploits: Will Eatherton on AI-Driven Security and the End of Obscurity
40mAI As The New Hacker Frontier
57mAI as the New Hacker Frontier
05mThe Asymmetry of AI Security and Physical Hardware Resilience