OpenAI launches GPT 5.4 Cyber to boost AI-powered cybersecurity, eyes lead over Claude Mythos

OpenAI, on March 14, announced to expand its Trusted Access for Cyber (TAC) program with the launch of a new GPT 5.4 Cyber model, a dedicated variant of GPT-5.4. The model is specially designed to boost AI-powered cyber defences and to empower security professionals to do advanced defensive work. 

The announcement comes a week after Anthropic launched Claude Mythos under Project Glasswing, showcasing the growing AI arms race in cybersecurity, as tech giants rapidly develop more powerful tools to detect vulnerabilities, counter emerging threats, and strengthen digital defences. 

What makes GPT 5.4 Cyber powerful is the binary reverse engineering, which allows experts to examine software, detect malware risks, find security vulnerabilities, and assess how strong or secure the software is, without needing the source code. Here’s everything you need to know about GPT 5.4 Cyber.

Also read: BT explainer: Anthropic’s Claude Mythos preview is here to reshape cybersecurity—here’s how

What is GPT 5.4 Cyber?

The GPT‑5.4‑Cyber is a model that is fine-tuned for AI-powered cyber defences, and comes with fewer capability restrictions. This is a tailored version of GPT‑5.4, which is willing to respond to cybersecurity-related requests when they are legitimate. 

The model is designed to support advanced defensive workflows such as identifying vulnerabilities in large codebases, reasoning across complex software systems, analysing malware behaviour, and binary reverse engineering without source code access

Since the model is less restrictive in what it allows, OpenAI is being cautious about how it’s released. The rollout of the model is restricted to trusted security companies, organisations, and researchers. The company stated that, “Access to permissive and cyber-capable models may come with limitations, especially around no-visibility uses like Zero-Data Retention⁠(opens in a new window) (ZDR).”

Also read: ‘Wrongdoers must be held accountable,’ says Florida AG as probe hits OpenAI amid IPO plans

OpenAI GPT 5.4 Cyber vs Anthropic’s Claude Mythos

While both AI models are developed for cybersecurity purposes, the GPT 5.4 is a tailored version of the GPT 5.4 model. In contrast, Anthropic’s Claude Mythos is developed from scratch and is an entirely new model.

Mythos is available in preview to only a few organisations under the Project Glasswing to test for cyber defence, and is not available for general public release. On the other hand, GPT 5.4 Cyber is part of a controlled access under its TAC program, which was announced back in February 2026. This programme enables organisations and researchers to access restricted models after strict identity verification.