OpenAI latest model GPT-5.4 Cyber has been launched to help security teams detect vulnerabilities in software faster and strengthen digital defenses against cyber threats.
OpenAI has introduced GPT-5.4 Cyber, a cybersecurity-focused AI model designed to help defenders identify and fix software vulnerabilities more quickly. OpenAI latest model marks a growing effort among major AI companies to apply advanced models to digital defense while managing the cyber risks.
Alongside the launch, OpenAI said it is expanding its Trusted Access for Cyber (TAC) program to thousands of verified individual security professionals and hundreds of teams responsible for protecting critical systems. The company said AI powered cybersecurity can significantly speed up defensive work by helping experts detect issues in complex software environments.
OpenAI said its approach focuses on broad access with controlled safeguards. It is launching the GPT-5.4 Cyber model gradually while improving protections against misuse, including jailbreak attempts and adversarial prompts. Its goal is to support responsible use at scale while strengthening defenses as model capabilities advance.
The release comes shortly after the preview of Anthropic's Mythos, a frontier model that's being deployed as part of Project Glasswing aimed at identifying software weaknesses. OpenAI also pointed to its Codex Security system, which has already helped discover and fix more than 3,000 critical and high severity vulnerabilities in software systems.
However, OpenAI also warned that AI systems are dual-use. This means the same capabilities that help defenders can also be used by attackers. The company noted a concern that adversaries could adapt security-trained models to discover and exploit vulnerabilities in widely used software before patches are available. This could increase risks for users and organizations if safeguards are not strong enough.
The company said it will continue working with security researchers and industry partners to improve model safety, strengthen detection systems, and ensure that defensive AI tools remain reliable as cyber threats evolve over time in increasingly complex digital environments worldwide across critical infrastructure and enterprise networks globally in real time.
According to Business Fortune, as cybersecurity and AI work together, developments like GPT-5.4 Cyber signal a new phase in digital defense, where speed, scale, and responsibility will define the future of security.
.webp)
