OpenAI launches tools to automate web app security testing and patching

Automating the security workflow

OpenAI recently announced two new cybersecurity products, Daybreak Red Team and Daybreak Defense Factory, during its Cyber at the Frontier event. These tools are designed to help organizations manage the complex process of finding, validating, and fixing security flaws in web applications. As reported by CyberSecurityNews, the launch aims to bridge the gap between identifying a potential weakness and actually repairing it.

The primary goal of these tools is to solve a common problem in cybersecurity where teams are overwhelmed by alerts. Simply finding a possible vulnerability is only the first step in a long process. Security teams must still determine if an attacker can actually reach the flaw, what impact it would have, and whether a proposed fix truly removes the risk without breaking other parts of the software.

By linking automated penetration testing with vulnerability remediation, OpenAI is attempting to create a more streamlined workflow. The Daybreak platform focuses on providing practical evidence rather than just increasing the volume of reported issues. This approach ensures that security researchers and developers can prioritize the most critical threats first.

How Daybreak Red Team works

Daybreak Red Team is currently available in early access and uses OpenAI models to scrutinize web applications for security risks. Unlike a standard scanner that might flag every possible error, this tool uses AI agents to investigate the specific context of an application. It looks at the source code and possible attack paths to see if a reported issue presents a real danger.

The tool is versatile enough to handle various types of input from different sources. It can examine findings submitted through bug bounty programs, vulnerability disclosure reports, or existing code scanners. By analyzing these inputs, the system can verify whether a flaw is reachable and exploitable in a real-world environment. This helps teams filter out noise and focus on legitimate threats.

Nancy Wang, who leads cyber-applied AI at OpenAI, explained that the service combines continuous scanning with deep investigation. The goal is to move past a simple warning and provide a clear picture of the risk. This allows organizations to understand the severity of a bug before they commit resources to fixing it.

From discovery to code patches

While Red Team identifies the problems, Daybreak Defense Factory is responsible for creating the solutions. This tool uses specialized models, including GPT-6 Cyber, to investigate potential flaws across entire codebases. Once a vulnerability is confirmed, the AI agents work to develop proposed patches to resolve the issue.

One of the key features of Defense Factory is that it provides engineers with a merge-ready pull request. This means the tool doesn't just suggest a fix in a vacuum; it provides the actual code change along with a test showing that the fix works. This gives developers the evidence they need to assess the change during a standard code review process.

This functionality makes the output significantly more useful than a standalone alert. Instead of a developer having to figure out how to fix a bug from scratch, they are presented with a ready-to-use solution. This helps security researchers and developers resolve vulnerabilities much faster than manual methods allow.

Integration into development

These new tools build upon OpenAI's previous work with Codex Security and Codex Security Cloud. Those earlier projects established a foundation for building project-specific security context and validating findings in isolated environments. The new Daybreak products extend this approach into a more continuous security model.

OpenAI is moving toward a system of continuous security checks that live within the existing development workflow. This includes monitoring repositories and incoming commits to identify issues as they are written. By removing duplicate findings and preparing fixes automatically, the system helps maintain a steady state of security without constant manual intervention.

The platform is designed for approved defenders who work within established security and development pipelines. It aims to provide the necessary alert context to stop incidents quickly. By integrating these checks into the daily work of engineers, the tools help ensure that security is a constant part of the software lifecycle.

Human oversight and limitations

Despite the high level of automation, OpenAI has maintained a clear boundary regarding human involvement. The Defense Factory prepares changes rather than deploying them independently. This ensures that human engineers remain the final authority on what code is allowed to enter a production environment.

The practical value of these tools will depend on the reliability of the evidence and the quality of the patches provided. If the AI produces accurate results, it can significantly reduce the time spent on lengthy manual investigations. However, because Daybreak Red Team is currently in early access, its immediate availability and performance may still be evolving.

OpenAI's strategy is to complete a full workflow from discovery to verification. By providing both the identification of a weakness and the proposed fix, they are positioning these tools as a comprehensive suite for modern security teams. The focus remains on providing useful, actionable data that helps teams secure their applications more efficiently.