OpenAI Debuts Codex Security for Automated Vulnerability Detection

OpenAI launched Codex Security on March 7, 2026, introducing an AI-powered security agent designed to automatically discover, validate, and suggest fixes for code vulnerabilities. The company positioned this as a research preview exclusively for its premium customer tiers.

The new security agent operates through the Codex web platform, building comprehensive context about development projects to identify potential security weaknesses. OpenAI structured the rollout as a gradual deployment to enterprise and educational customers first.

Premium ChatGPT Customers Get Early Access

The Codex Security agent targets ChatGPT Pro, Enterprise, Business, and Education subscribers exclusively. Individual ChatGPT users and free-tier customers won't have access during this research preview phase.

Enterprise development teams and educational institutions with existing ChatGPT subscriptions can immediately access the security features through their Codex web interface. OpenAI hasn't announced timeline for broader availability.

Related: Meta Deploys AI-Powered Anti-Scam Tools Across Platforms

Related: Betterleaks Tool Scans Git Repos for Exposed Secrets

Related: WhatsApp Launches Parent-Managed Accounts for Pre-Teens

Related: Meta Acquires Moltbook AI Agent Social Network

Related: OpenClaw AI Critical RCE Flaw Patched — All Developers Must

Free Trial Period and Implementation Details

OpenAI offers free usage of Codex Security for the first month following activation. The AI agent analyzes project codebases to build contextual understanding before flagging potential vulnerabilities and recommending remediation steps.

Organizations can access the feature directly through their existing Codex web dashboard. The CISA Known Exploited Vulnerabilities catalog provides additional context for security teams evaluating AI-suggested fixes against known threat patterns.