OpenAI met with 50 cyber defense practitioners from federal agencies, state governments, and Five Eyes allies to present its GPT-5.4-Cyber model. The model launches through a dual-track approach with a broadly accessible version and a separate version reserved for defenders through its Trusted Access program. The release follows an intelligence-sharing partnership with Australia, Canada, New Zealand, and the U.K. amid warnings that AI tools could be misused to disrupt critical infrastructure.
