OpenAI has launched a new AI agent, Codex Security, designed to identify and fix cybersecurity threats. The company highlights that this tool can handle vast amounts of data and generate precise patches, allowing developers to focus on more complex tasks.
Codex Security is built on the Aardvark project, which utilized the GPT-5 model that OpenAI tested last year in a closed environment. The AI now serves as a commercial product, capable of scanning repositories and learning from user feedback to enhance its vulnerability detection algorithms.
The introduction of Codex Security intensifies competition in the cybersecurity industry, challenging traditional security firms and contributing to the ongoing trend of automating security solutions. This follows the earlier release of Claude Code Security by Anthropic, which sparked a noticeable drop in stock prices of companies like Crowdstrike and Cloudflare by 8%.
Codex Security will be available in the coming days as an early preview for corporate clients using ChatGPT Enterprise and Business, as well as educational institutions. A free one-month trial will be offered to allow organizations to evaluate the system's effectiveness without any upfront costs.
This move signals a growing push towards AI-driven cybersecurity solutions, which could disrupt the market and prompt competitors to accelerate their own innovations.
Informational material. 18+.
