OpenAI Launches GPT-5.4-Cyber for Cybersecurity Professionals

ME News reports that on April 15 (UTC+8), according to monitoring by Beating, OpenAI has released GPT-5.4-Cyber, a model variant fine-tuned specifically for cybersecurity scenarios based on GPT-5.4. The key distinction from the standard version is its actively lowered refusal threshold for security-related requests, enabling security professionals to perform operations that conventional models would typically reject—such as binary reverse engineering, which involves analyzing compiled software for malicious code or vulnerabilities without access to the source code. This model is not available to all users but is deployed in tiers through OpenAI’s Trusted Access for Cyber (TAC) program. GPT-5.4-Cyber is accessible only to the highest tier of users: vetted security vendors, enterprise security teams, and cybersecurity researchers. OpenAI also noted that users employing this more permissive model may face additional restrictions, particularly in low-visibility scenarios such as Zero Data Retention (ZDR), with stricter controls when accessing via third-party platforms. The TAC program itself is expanding and has now grown to include thousands of individual defenders and hundreds of security teams. Individual users can verify their identity at chatgpt.com/cyber to gain basic-level reduced-refusal access, while enterprises must apply through an OpenAI account representative. OpenAI states that this tiered strategy will guide the testing and deployment of future, more powerful models: standard models with general safety measures will be broadly released, while specialized, permissive models trained for cybersecurity will always be deployed in a restricted manner. (Source: BlockBeats)