OpenAI announces 'Daybreak,' a security AI that detects and fixes vulnerabilities before attackers can.

OpenAI, the developer of ChatGPT, has announced Daybreak , a security AI. Daybreak is a cutting-edge security AI that can detect and fix vulnerabilities in systems and software before attackers can exploit them, and is attracting attention as an AI that can counter Anthropic's Claude Mythos .

Daybreak | OpenAI for cybersecurity | OpenAI
https://openai.com/daybreak/

OpenAI just released its answer to Claude Mythos | The Verge
https://www.theverge.com/ai-artificial-intelligence/928342/openai-daybreak-security-ai

OpenAI explained that next-generation cybersecurity and Daybreak 'start from the premise that vulnerability resilience should be built into software from the design stage, rather than just discovering and fixing vulnerabilities. AI helps defenders analyze codebases across the board, identify vulnerabilities, verify fixes, analyze unfamiliar systems, and accelerate the discovery-to-remediation process. However, these capabilities can also be exploited, so Daybreak combines enhanced defensive capabilities with reliability, verification, appropriate security measures, and accountability.'

Daybreak's goal is to accelerate cybersecurity and ensure the continuous security of software. By combining the advanced intelligence of OpenAI's AI models, the extensibility of Codex's agent-based tools, and collaboration with partners across the entire security field, Daybreak makes it possible to create a safer world for everyone.

Specifically, Daybreak combines GPT-5.5-Cyber , an AI model for security that OpenAI just released on May 6, 2026, with Codex Security, an AI agent released in March 2026. Codex Security builds editable threat models from repositories and analyzes them focusing on realistic attack vectors and high-impact code.

OpenAI announces 'Codex Security,' an AI agent that automates vulnerability discovery, verification, and remediation - GIGAZINE

Daybreak allows you to integrate secure code reviews, threat modeling, patch validation, dependency risk analysis, malware detection, and remediation guidance into your daily development process. This enables you to create more robust software from the moment of release.

OpenAI CEO Sam Altman commented on Daybreak, saying, 'OpenAI is launching Daybreak, an initiative to accelerate cyber defense and continuously protect software. AI is already good at cybersecurity and will soon be even better. We now want to work with as many companies as possible and help them continuously protect themselves.'

OpenAI is launching Daybreak, our effort to accelerate cyber defense and continuously secure software.

AI is already good and about to get super good at cybersecurity; we'd like to start working with as many companies as possible now to help them continuously secure themselves.

— Sam Altman (@sama) May 11, 2026

Over the next few weeks, OpenAI will work with industry and government partners to prepare for the deployment of a model with progressively increased cybersecurity readiness as part of an iterative deployment approach.

Additionally, a vulnerability scan request form is available for companies that wish to request a vulnerability scan from Daybreak.

Request a vulnerability scan | Daybreak | OpenAI
https://openai.com/daybreak/request-a-vulnerability-scan/