OpenAI Launches Daybreak, an AI Cybersecurity Rival to Claude Mythos

OpenAI has officially launched Daybreak, an artificial intelligence (AI)-based cybersecurity initiative, on Monday (11 May 2026). Daybreak is designed to assist companies in automatically detecting and repairing various security vulnerabilities in software. The introduction of Daybreak is regarded as OpenAI’s response to its rival Anthropic’s AI cybersecurity technology, Claude Mythos. Recently, Claude Mythos has been in the spotlight for its claimed ability to identify security vulnerabilities faster than humans. On the other hand, the technology has also sparked controversy because it is said to be capable of discovering zero-day vulnerabilities unknown to developers, raising fears that it could be misused for hacking activities. OpenAI explains that Daybreak is built with the vision that cybersecurity should not only focus on finding and patching vulnerabilities after they are discovered. According to OpenAI, security systems need to be designed from the outset to make software more resilient to cyber attacks. “AI can now help defenders understand the codebase, identify hidden vulnerabilities, validate fixes, analyse unknown systems, and accelerate the process from issue discovery to remediation,” OpenAI states. In practice, Daybreak is claimed to help security teams prioritise the most dangerous threats and reduce the analysis process from hours to minutes. The system can also create and test security patches directly in the company’s code repositories. Subsequently, OpenAI’s system will build a threat model from those code repositories via Codex Security. After that, Daybreak will focus analysis on realistic attack paths and high-risk code. In addition to testing attack threats, Daybreak is also claimed to be able to deliver analysis results along with audit evidence to the company’s systems to aid in the verification and remediation process.