AI Hackers Are Faster Than Your Pen Test — Eli Cohen, Snyk
- Source
- youtube.com
- Author
- AI Engineer
- Date
Context
Eli Cohen of Snyk, a security company, argues that yearly penetration tests cannot keep up with attackers who use AI. A penetration test is a hired attempt to break into a system to find weaknesses. In his AI Engineer World's Fair 2026 talk, as described, he says attackers use frontier models to chain small vulnerabilities into critical ones in minutes, citing a fastest AI attack of four minutes, while human pen tests happen once or twice a year.
His proposed answer is continuous offensive security: AI pen testing on every code change, run by a team of agents covering orchestration, reconnaissance, vulnerability hunting, an exploit-validating judge, remediation and reporting. The judge step matters because proving an exploit works is how the approach aims to cut false positives.
The four-minute figure and the critique of older tools are the speaker's claims, and the description gives no measurements of how well the approach finds real flaws.
- Agent — A system in which a model repeatedly chooses operations, a harness carries them out, and the results return to the model to inform its next choice.
Checking sign-in…
Loading comments…






