Daybreak Is OpenAI's Response To Anthropic's Claude Mythos

OpenAI launched Daybreak, a cybersecurity initiative positioned as a rival to Anthropic’s Project Glasswing.

The program uses OpenAI models, including Codex Security and GPT-5.5 variants, for code review, vulnerability triage, malware analysis, detection engineering, patch generation, and validation.

OpenAI says Daybreak is meant to build security earlier into software development and produce audit-ready evidence.

The company is partnering with firms including Cloudflare, Cisco, CrowdStrike, Palo Alto Networks, Oracle, and Akamai.