Democratizing Security for the Agentic Web with Cloudflare CTO Dane Knecht

Cloudflare CTO Dane Knecht argues that AI has turned security from an afterthought into the main engineering priority, with the attacker landscape shifting from occasional landmark exploits to a much faster cadence of vulnerability discovery. He says the next 18 months will likely see most engineering time redirected toward security work, not because product development stops, but because organizations have to patch a huge backlog while also hardening new AI systems. He expects the industry to spend more on red teaming, defensive review, and process changes that make security part of the software and agent development lifecycle.

Knecht is less apocalyptic than people calling for air gaps everywhere. He says companies that already moved from castle-and-moat thinking to zero trust are in better shape, and he thinks humans will adapt when critical infrastructure is at stake. He also argues that AI should reduce complexity, not add it: simplify code, shrink attack surface, use safer languages and formal methods where possible, and embed security checks into the workflow. At Cloudflare, that means more red teams, more internal standards, and agents that follow documented rules.

He also lays out Cloudflare’s view of the agentic web as a “messy middle” where humans and agents share the same rails. That pushes the company toward new payment and discovery infrastructure, including microtransactions for content, agent wallets, and tools that help sites monetize when agents consume their pages. Internally, Cloudflare OS is meant to give employees a secure way to use AI while teaching agents how Cloudflare works; Knecht says the company is still early, but the opportunity is to make every team faster and more autonomous.