White House Limits GPT 5.6 Release for Security Reasons

OpenAI's new GPT 5.6 model is not being immediately released to the public. The Trump administration asked OpenAI to restrict early access for security reasons. OpenAI plans to offer GPT 5.6 only to a select group of partners during a preview period. OpenAI CEO Sam Altman stated the government will approve customer access one by one. If the trial phase goes well, a wider public launch could happen a few weeks later. The Office of the National Cyber Director and the Office of Science and Technology Policy reportedly requested this limited launch. The government is also reviewing the new model, and OpenAI has worked closely with various agencies. This follows the Trump administration's push for federal oversight of new AI models. Earlier this month, an executive order required some AI companies to submit models for testing before public release. Similar concerns arose with Anthropic's Claude Mythos cyber model, which was also limited to a small group due to potential misuse. Advanced AI models, especially large language models, raise cybersecurity worries. They can create malware and even carry out ransomware attacks. Cyber models like Claude Mythos might also find and exploit software vulnerabilities at high speeds, potentially increasing cyber attack risks. The actual threat is still hard to measure since these models are not yet public. This matters because it highlights growing government concern over the security implications of powerful new AI technologies.