Building an AI-Powered Defense Factory with OpenAI's Codex Security

OpenAI's cyber product engineering team, Ian and Kyle, present Codex Security, a set of tools for finding, validating, and fixing software vulnerabilities with AI. They share lessons from OpenAI's internal security sprint and show how the same workflows can be applied to your own codebase.

■ Internal Security Practices
- The current cybersecurity landscape and the concept of the defender's window
- How OpenAI ran an internal security sprint and built a continuous "defense factory" loop

■ Codex Security CLI Workflows
- Giving agents context: custom scan instructions, knowledge bases, and security .md files
- Running deep scans in parallel and deduplicating findings
- Enriching findings with a security rubric and ownership assignment
- Generating and verifying patches, then integrating them into your SDLC

■ Codex Security Cloud
- A turnkey version of the CLI workflows with continuous scanning
- Powered by Daybreak-class models, installed via the plugin marketplace

This session is aimed at developers and security engineers who want to reduce vulnerability backlogs and adopt AI-assisted security practices. Viewers can learn the end-to-end workflow, see a live demo, and take concrete steps such as installing Codex Security from npm or setting up Codex Security Cloud with their GitHub repositories.