OpenAI Launches Codex Security SDK for AI-Powered Vulnerability Scanning

OpenAI has released Codex Security, a new CLI and TypeScript SDK that brings LLM-powered vulnerability detection to developers. The tool uses GPT-5.6-terra to scan, validate, and fix security issues directly in your codebase, marking OpenAI's entry into the automated security tooling space.

Codex Security runs scans via a simple npx command and supports both interactive use through ChatGPT credentials and CI integration with an API key. It requires Node.js 22 and Python 3.10 or later, making it straightforward to integrate into existing development workflows.

For founders and developers building AI products, this launch signals a shift toward LLM-native security tooling. Unlike traditional pattern-matching scanners, Codex Security can reason about code context and detect logic-level vulnerabilities that rule-based systems might miss.