OpenAI Launches Daybreak, Two AI Tools for Cybersecurity

OpenAI has launched two cybersecurity tools under the Daybreak brand. Powered by GPT-6 Cyber, they aim to automate both code analysis and patch production, with pull requests ready for review. The offensive testing component, Daybreak Red Team, is offered in early access.

Defense Factory automates patches and delivers pull requests

Daybreak Defense Factory automates defense by analyzing a company's code for vulnerabilities and generating patches. The tool provides pull requests ready to be deployed, accompanied by fixes and tests, for approval by developers. Both Daybreak products operate with OpenAI's GPT-6 Cyber model. This launch introduces a defensive workflow under the Daybreak brand that extends to proposing code modifications.

Red Team industrializes vulnerability research with AI agents

Daybreak Red Team, available in preview, uses artificial intelligence agents to automatically examine web applications for vulnerabilities. The tool reviews the source code as well as the application context to map out attack vectors, checks vulnerabilities identified by bug bounty programs or code analysis tools, and tests the patches applied. Nancy Wang, who leads applied AI in cybersecurity at OpenAI, states that the solution aims to provide penetration testing at a pace equivalent to that of continuous automated scanning while achieving the depth of human experts.