OpenAI responds to Anthropic with a more accessible cybersecurity program

OpenAI has launched Daybreak, a cybersecurity initiative positioned as a more open and less ominous counterpart to Anthropic’s Project Glasswing. Like Glasswing, Daybreak partners with companies and governments to detect vulnerabilities using frontier AI models, but OpenAI emphasizes accessibility, offering public‑facing tools such as a simple vulnerability‑scan request form. Daybreak uses Codex Security to build threat models, map trusted components, and analyze real codebases for exploitable weaknesses. Unlike Anthropic’s restricted Mythos‑based program, OpenAI is actively encouraging broad adoption, framing Daybreak as a practical, transparent way to strengthen software from the ground up.

Read the full story on Gizmodo →