Cloudflare uses OpenAI models to prioritize customer vulnerabilities
Original title: Cloudflare links vulnerability discovery to defense with OpenAI
The early-access service proposes code patches and custom WAF rules after customers authorize code review.
Cloudflare is offering Vulnerability Discovery and Remediation through Managed Defense, using OpenAI models accessed through Daybreak to inspect customer-authorized code. Cloudflare adds live traffic, security signals, and validation, then prepares proposed patches and, where appropriate, custom web application firewall rules. Cloudflare says it runs automated checks and reviews proposals before customers decide what to implement. Separately, Cloudflare says its developers use Codex, and Mark Anderson described a four-person rebuild finished in three days.
Unconfirmed: The productivity, staffing, and token-cost figures are estimates stated by Cloudflare's Mark Anderson.
- The service is early access through Cloudflare Managed Defense.
- Findings are tied to whether a route is active or seeing suspicious traffic.
- Anderson says every Cloudflare software developer uses Codex.
- He estimates the rebuild cost $1,100 in tokens and would otherwise have needed far more developers.
- Vendors
- CloudflareOpenAI
From the original article
3 min read at openai.com
Cloudflare uses OpenAI Daybreak to investigate vulnerabilities and propose fixes for review. Its developers use Codex to build faster and improve code quality.
openai.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.