# OpenAI Launches Codex Security to Automatically Find and Fix Code Vulnerabilities

OpenAI announced today that Codex Security, an AI-powered application security agent, is now available in research preview.

The new tool goes beyond traditional security scanners by analyzing entire project contexts to detect vulnerabilities, validate whether they're genuine threats, and automatically generate patches to fix them. According to OpenAI's announcement, this approach delivers "higher confidence and less noise" compared to existing security tools.

This matters because software security remains a critical challenge for development teams. Traditional security scanners often overwhelm developers with false positives—flagged issues that aren't actually exploitable. By understanding project context and validating vulnerabilities before alerting teams, Codex Security could significantly reduce wasted time investigating non-issues while catching real threats that matter.

The automatic patching capability is particularly noteworthy. Rather than simply identifying problems and leaving developers to fix