AI for Cybersecurity Teams | OpenAI

OpenAI

2 min read Original article ↗
OpenAI

Watch the recording:

See how frontier AI is reshaping cybersecurity

Watch recording

Models for defense and red teaming

Use Daybreak models with Codex Security or your own security harness.

Designed for security work at scale

  • 30M+

    commits scanned

  • 30K+

    codebases scanned

  • 500K+

    findings fixed

Expand the capacity of every security team

Codex Security, wherever your teams work

Use Codex Security across the tools your teams already use. Scan code in the cloud, review pull requests automatically, or run local security checks from your terminal with the open-source Codex Security CLI.

Three-circle icon representing the Codex Security CLI.

Codex Security CLI

Best for running security scans from your terminal and integrating them into local development or CI/CD workflows.

Copy the following code, go to Codex and bash in your CLI: curl -fsSL https://openai.com/codex/security/scan.sh | bash

Built for the full security workflow

Find vulnerabilities in code

Codex Security + Daybreak Blue: Scan repositories and code changes for likely vulnerabilities, using codebase context to surface high-signal findings before they reach production.

Read the documentation(opens in a new window)

Validate what matters

Codex Security + Daybreak Blue: Reproduce candidate findings, establish reachability and impact, and capture evidence so teams can focus on vulnerabilities that present real risk.

Read the documentation(opens in a new window)

Anticipate how systems may be attacked

Daybreak Blue: Map assets, entry points, trust boundaries, sensitive data paths, and abuse cases to help teams prioritize security work before incidents occur.

Read the documentation(opens in a new window)

Test systems like an adversary

Daybreak Red: Conduct scoped penetration tests and realistic red-team exercises to validate defenses and demonstrate how vulnerabilities could be chained, under clear authorization and rules of engagement.

Read the documentation(opens in a new window)

Investigate complex security targets

Daybreak Red: Analyze malware, binaries, firmware, and controlled proofs of concept to investigate bounded security hypotheses when source code is incomplete or unavailable.

Read the documentation(opens in a new window)

Trusted by leading security organizations

Choose how to get started

Start with Codex Security, request advanced access, or work with us to bring Daybreak capabilities to your customers.

Frequently asked questions