OpenAI has introduced a new specialized artificial intelligence model named GPT-5.4-Cyber to enhance cybersecurity and protect digital infrastructure. Announced on April 24, 2026, the model is designed to help security experts quickly identify and fix vulnerabilities in systems, making data and user protection more efficient. The initiative underscores OpenAI’s growing focus on cyber defense amid rising digital threats.

Built on the company’s flagship GPT-5.4 model, GPT-5.4-Cyber aims to support real-time detection and prevention of cyberattacks. To prevent misuse, OpenAI has expanded its Trusted Access for Cyber (TAC) program, allowing thousands of authorized security professionals and hundreds of teams to use the advanced AI tool. The company described this as an example of democratizing access to cutting-edge technology.

OpenAI acknowledged that while AI can strengthen defenses, it can also be exploited by hackers. The firm has therefore added strict safeguards to ensure vulnerabilities are patched before criminals can exploit them. The launch follows Anthropic’s release of its Mythos model and builds on OpenAI’s earlier success with its Codex Security tool, which has resolved over 3,000 major security flaws.