October 21, 2026
Cloudflare combines real-time traffic and security signals with OpenAI Daybreak to investigate vulnerabilities and propose fixes for customer review.

Company size: Enterprise
Region: North America
Industry: Technology
Products: Daybreak, Codex
234 billion
Cyber threats blocked daily on average by Cloudflare’s network in Q1 2026
10x
Reduction in development resources required to build a new product
3 days
Time taken to refactor a product using Codex
$1,100
Token spend for the product refactoring
Cloudflare operates one of the world’s largest networks, proxying approximately 20% of global web traffic. Organizations rely on it to keep their applications and services secure, fast, and reliable.
In its early-access service offered through Cloudflare Managed Defense—Vulnerability Discovery and Remediation—Cloudflare pairs OpenAI models accessed via Daybreak with its global network and threat context. These models review code and assess potential vulnerabilities. Cloudflare then combines real-time traffic data, security signals, and verification mechanisms to prioritize findings and draft proposed mitigations and patches for customer review.
Codex further expands OpenAI’s footprint across Cloudflare’s engineering teams, where developers leverage it to boost output and improve code quality.
“The beauty of the automation brought by Daybreak and Codex is that you can let machines handle the heavy lifting, freeing humans to focus on the most critical engineering and security decisions.”
— Mark Anderson, President of Revenue, Cloudflare
Scaling security analysis with Daybreak
Security teams face an endless stream of vulnerability reports and alerts every day. Sifting through this data manually is both time-consuming and labor-intensive, leaving teams with insufficient bandwidth to investigate and address critical issues. “Traditional security teams sit on massive amounts of useful data—they just lack the people and time to process it,” says Anderson.
Cloudflare’s approach correlates vulnerabilities in source code with what is actively happening in production environments. Is the affected route live? Is it receiving suspicious traffic? Are existing security controls already protecting it? These signals provide security teams with richer context to help determine which issues to prioritize for investigation.
Through Vulnerability Discovery and Remediation, Cloudflare uses OpenAI models accessed via Daybreak to review customer-authorized code, identify potential vulnerabilities, and verify the evidence supporting each finding. The service prepares proposed code patches and, where appropriate, provides custom Web Application Firewall (WAF) rules to reduce exposure while developers review the fixes. Cloudflare runs automated checks on these proposals, and its team conducts human reviews before sharing recommendations with customers.
For Cloudflare customers, the benefit lies in a clear path from discovering a vulnerability to protecting the affected application. The service provides rationale for prioritizing vulnerabilities, edge protection recommendations to mitigate exposure risk, and code fixes for review—all while ensuring customers retain full control over which changes are ultimately deployed.
Refactoring a product in 3 days with Codex
Securing applications also depends on the code teams write and maintain. Anderson links Cloudflare’s use of Codex directly to higher developer productivity, improved code quality, and fewer defects introduced during development.
“At Cloudflare, a third of the company are software developers building services and features for our platform, and 100% of them are using Codex to increase output and ship better code.”
— Mark Anderson, President of Revenue, Cloudflare
Anderson described a refactoring project the team had previously deferred. “We kept putting off building it because we just didn’t have the people to do it,” he said. With Codex dramatically boosting developer productivity, the question facing the team shifted from “Can we take this project on?” to “Why wouldn’t we?”
Anderson estimated that the effort would have normally required months of roadmap planning and “likely up to 10 times as many developers.” Powered by Codex, a team of four Cloudflare developers took on the build. “We built it, deployed it, and tested it in three days, with just $1,100 in token costs,” he said.
Building cyber defense around AI and human judgment
For Anderson, integrating AI into everyday engineering means “reinventing internal processes and building technology with machines in mind—not as isolated elements, but as deeply integrated parts of the development workflow.”
Cloudflare’s early-access security service puts this philosophy into practice by pairing code reviews with production evidence and proposed actions for customer evaluation. The goal is to equip security teams with the information needed to make critical decisions about which vulnerabilities to prioritize, and how to protect and remediate affected applications—while keeping customers firmly in control of every change.