🤝 What Happens When Attackers Can Automate Exploits? Inside Maze's AI-Native Security Defense

🤝 What Happens When Attackers Can Automate Exploits? Inside Maze's AI-Native Security Defense
Theory Ventures' Bryan Bischof interviews Harry Wetherald, co-founder and CEO of Maze, about AI agents that continuously investigate cloud vulnerabilities and help developers fix them before attackers exploit them. Once an exploit works, attackers automate and repeat it for pennies per run, so defenders now need systems that catch anything, not just known CVEs. Maze evolved from a single all-powerful agent to hundreds of thousands of narrowly scoped micro-agents, adding permission boundaries once it understood what access each agent actually needed. The real challenge isn't detection accuracy - it's making that accuracy affordable at scale, a cost dynamic increasingly relevant to fintech and other regulated industries.
🔗 Read more