๐ค Theory Ventures: Inside Maze's Fight Against AI-Automated Cyberattacks
In a Theory Ventures conversation, Maze co-founder and CEO Harry Wetherald explains how AI now lets attackers turn a working exploit into an automated, repeatable attack chain at the click of a button, forcing defenders to hunt for unknown vulnerabilities rather than just known CVEs. Maze's platform, which plugs into a company's cloud environment to investigate and help fix vulnerabilities, evolved from one broad AI agent into hundreds of thousands of narrowly scoped micro-agents, each given minimal permissions to keep risk contained. Wetherald says the hardest engineering problem isn't raw accuracy but accuracy at affordable scale, since naive approaches can cost millions in inference to investigate millions of vulnerabilities โ underscoring continued VC appetite for AI-native cybersecurity defense.
๐ Read more