OpenAI Unveils GPT-5.4 Cyber: A Direct Challenge to Anthropic

OpenAI Unveils GPT-5.4-Cyber: A Direct Challenge to Anthropic

OpenAI has recently introduced a new version of its artificial intelligence model, ChatGPT, specifically focused on cybersecurity. This model, named GPT-5.4-Cyber, is a specialized extension of GPT-5.4, which was launched shortly before. This version aims to enhance cybersecurity defense capabilities, but it is not intended for general use. Only researchers and industry experts will have access to it, according to OpenAI's statements.

This announcement comes at a time when Anthropic has recently launched Claude Mythos, an AI designed to identify vulnerabilities and simulate exploitation scenarios. The proximity of these announcements raises questions about a possible direct competition between the two entities.

The Capabilities of GPT-5.4-Cyber

One of the main strengths of GPT-5.4-Cyber lies in its ability to analyze compiled software without requiring access to the source code. This means that the AI can:

• Examine unknown programs
• Identify suspicious behaviors
• Detect vulnerabilities
• Assess the robustness of a system

Additionally, it is capable of dissecting malware and spyware to understand their flaws, a task that standard versions of ChatGPT cannot perform. OpenAI has also indicated that this model will serve as a foundation for future developments of their cybersecurity tools, with even more advanced versions already in the works.

Restricted Access and Controlled Use

To prevent any misuse, OpenAI has decided to restrict access to GPT-5.4-Cyber. Only organizations that can demonstrate their legitimacy will be able to use this model. These include:

• Companies specializing in cybersecurity
• Teams responsible for protecting critical infrastructures

To obtain the necessary permissions, a verification system will be implemented. OpenAI has introduced a program called Trusted Access for Cyber (TAC), which operates on levels. The more verified an entity is, the more advanced features it can unlock.

The highest access levels will allow the model to be utilized in its most powerful version, capable of competing with a human expert in vulnerability detection. OpenAI plans to gradually expand this system to a broader network of professionals and specialized teams.

Unlike Anthropic, which restricts its Claude Mythos model to a few large tech companies such as Amazon, Google, or Microsoft, OpenAI advocates for a less centralized access model, allowing a greater number of qualified professionals to benefit from these advanced tools.