OpenAI gives certain users access to a new model that's better at finding software security vulnerabilities. Just a week ago, competitor Anthropic PBC announced the small-scale release of artificial intelligence tool Mythos.
The ChatGPT developer said on Tuesday it has begun rolling out the GPT-5.4-Cyber model, with the goal of identifying problems in the software so organizations can fix them. OpenAI says the model also has fewer restrictions on how users can detect it when used for such tasks. The model will be made available to select participants in OpenAI’s Trusted Access for Cyber program, which the company launched in February to allow select customers and cybersecurity professionals to try out its most powerful products.
OpenAI plans to expand participation in this early access program. Hundreds of users will be allowed to test the new model initially, expanding to thousands over the coming weeks.
OpenAI and Anthropic have been competing to develop more advanced artificial intelligence models with broad capabilities and trying to convince companies to pay for them. But as models become better at programming and code security detection, concerns are rising about the technology being misused by criminals and state-sponsored hackers. Artificial intelligence technology is currently being used to assist in cyberattacks.
Last week, Anthropic launched an artificial intelligence model called Mythos, saying it focuses on identifying and exploiting vulnerabilities in operating systems and web browsers. The company decided to release the model only to a small number of trusted partners, including Amazon, Apple and Microsoft, so that these organizations can use Mythos to detect problems and strengthen system protection.
