Cloudflare Vulnerability Discovery and Remediation Uses OpenAI Daybreak Models

What shipped

Cloudflare launched early access to Vulnerability Discovery and Remediation, a new invitation-only service inside Cloudflare Managed Defense. It uses OpenAI Daybreak models — including GPT-5.6 Cyber — through the OpenAI Daybreak Defense Network to hunt for and validate vulnerabilities in codebases the customer authorizes.

What changed

The service works across Cloudflare Workers and proxied apps. It snapshots traffic and security data from Web Assets and WAF, maps request paths to handling code, then sends hunter agents into authorized source. Findings must be corroborated by evidence before they surface. Cloudflare then proposes custom WAF mitigations and code patches, automatically checking each patch and mitigation before presenting them. The customer decides what gets deployed.

Why a builder cares

LLM scanners now flag thousands of issues in minutes — noise is the bottleneck. This ranks findings by production context: is the code live, on a hot path, under active attack, and unprotected. That is the difference between a generic finding and one you actually fix first, with patches validated before review.