Search
On April 14, 2026 (local time), OpenAI made an announcement that it would grant specific users access to its latest model, GPT-5.4-Cyber. This cutting-edge model boasts enhanced capabilities in detecting software security vulnerabilities, thereby streamlining the process for organizations to address and rectify these issues. Moreover, it offers greater flexibility in user probing methods, imposing fewer constraints. The model will initially be accessible to a select group of participants in OpenAI's 'Cybersecurity Trusted Access' program. This initial phase will involve hundreds of users testing the model, with plans to scale up to thousands in the subsequent weeks. Notably, just a week prior, its rival Anthropic had already rolled out its AI tool, Mythos, on a limited basis.
-
C114 Communication Network -
Communication Home
