OpenAI on May 11 introduced an AI-based cybersecurity initiative called Daybreak that uses AI to speed up identifying software vulnerabilities, verifying patches and strengthening security.

OpenAI CEO Sam Altman (샘 알트먼) said on X that Daybreak is an effort to accelerate cyber defence and keep software continuously safe. He said AI is already capable in cybersecurity and will soon become very powerful. He said he wants to work with as many companies as possible to help strengthen security on an ongoing basis.

Daybreak combines OpenAI AI models with Codex, an agentic system specialised in coding. Security teams can use it to review code, check external libraries that software depends on, model threats, verify patches and investigate unfamiliar systems or environments that are not internally understood. The goal is to cut the time from finding vulnerabilities to fixing them.

OpenAI stressed that AI can now help defenders analyse codebases, identify subtle vulnerabilities and move faster from discovery to fixes.

The announcement comes as concerns have risen about AI-based cyber attacks after Anthropic released Mithos last month. Mozilla said it used Mithos to find 271 previously unknown vulnerabilities in Firefox.

OpenAI plans to work with government and industry partners before deploying more powerful cyber AI models.