OpenAI has announced the launch of two new AI-powered cybersecurity tools, Daybreak Red Team and Daybreak Defense Factory, as part of its Daybreak security platform. The goal is to help companies identify and fix software vulnerabilities before hackers can exploit them.

According to "Indiatoday," the release comes amid growing concerns about AI-enabled cyberattacks. Hackers could potentially use more advanced tools to discover software weaknesses and automate attacks.

OpenAI believes that businesses need to reevaluate their protective measures to keep pace with these threats by leveraging AI to identify and address security vulnerabilities.

The two tools rely on AI agents to test applications, detect security weaknesses, investigate potential vulnerabilities, and assist developers in implementing necessary fixes before hackers can exploit them in cyberattacks.

Greg Brockman, CEO of OpenAI, said that the level of security companies have will depend on how advanced the AI models they use for defense are and how effectively they employ these models. He added, "We’re now at a point where companies won't be safe unless they have an advanced model to defend themselves and use it efficiently."

Brokmann noted that traditional security practices remain important, but stressed the need for institutions to rethink their protection methods in light of the capabilities provided by advanced AI models.

How does OpenAI’s new cybersecurity tool work?
 

Daybreak Red Team Tool

OpenAI designed the first tool, Daybreak Red Team, to test web applications and identify security vulnerabilities. It relies on a team of AI agents working in coordination to examine the application’s source code and understand its context, then search for potential ways that attackers might exploit it.

The tool functions like an automated penetration testing team; instead of waiting for hackers to discover vulnerabilities, companies can use it to simulate potential attacks and identify weaknesses themselves.

It can also analyze the outputs of code analysis tools, bug bounty programs, and vulnerability reports to verify whether reported issues are indeed real. After a company fixes any vulnerabilities, the tool can retest the application to ensure that the software update has successfully prevented the potential attack.

OpenAI allows companies to control which applications and endpoints AI agents are permitted to access. The tool uses isolated runtime environments and additional review mechanisms to ensure that tests do not exceed authorized scope. Daybreak Red Team is currently available through an early access program.

Daybreak Defense Factory tool

The second tool, Daybreak Defense Factory, focuses on addressing security vulnerabilities discovered by protection teams. It uses AI agents to examine the company's code, track down the source of the vulnerability, and prepare a software patch for developers to review before adoption. For example, if one of the security scanning systems detects a weakness in an application belonging to a company, the Defense Factory tool can investigate the issue and suggest a code modification to address it.

Developers can then review the evidence, test the proposed fix, and approve it for inclusion in the program. The tool can also handle results from existing security protection tools used by companies, along with those provided by Daybreak Red Team.

Companies can customize the tool to scan new code changes or perform periodic and scheduled scans as their applications are updated, helping to continuously identify and address vulnerabilities.