OpenAI launches Daybreak for AI security

OpenAI has introduced Daybreak, a new initiative designed to proactively identify and fix software vulnerabilities using AI. This program leverages the Codex Security AI agent to analyze code, model potential attack vectors, and automate the detection of high-risk security flaws. Daybreak's debut follows Anthropic's recent announcement of Claude Mythos, a security-focused AI model that Anthropic deemed too dangerous for public release, highlighting a growing trend in AI development for cybersecurity applications. OpenAI's move signals a competitive push in the AI security space, aiming to enhance software resilience against emerging threats.