OpenAI's ChatGPT-5.6 is subject to federal restrictions akin to those imposed on Anthropic’s Mythos model. According to a source, Washington has cautioned OpenAI against releasing the model without receiving approval. This indicates a growing trend of government oversight over advanced AI systems, particularly those with potential dual-use capabilities.

The U.S. government aims to ensure that its latest, most advanced AI tools cannot be used against national interests. OpenAI CEO Sam Altman mentioned during a staff Q&A that the latest model, GPT-5.6, is available in limited preview to a small group of customers selected by the U.S. government. This controlled release suggests a strategic approach to managing the risks associated with powerful AI systems.

The federal government, specifically the Office of the National Cyber Director and the Office of Science and Technology Policy, has requested that OpenAI stagger the release of its latest model. This approach is intended to mitigate potential security risks and ensure that the model is thoroughly evaluated before broader deployment.

Such restrictions may lead to increased costs for companies relying on AI tools, as well as potential delays in innovation. They could also result in greater vendor lock-in, as organizations may be forced to depend on specific models that are subject to regulatory approval. Additionally, these measures may influence market reactions, with some stakeholders viewing them as necessary safeguards and others as barriers to progress.

The situation remains fluid, with ongoing discussions between OpenAI and federal regulators. As the landscape of AI governance continues to evolve, the balance between innovation and security will be a critical factor in shaping the future of AI development and deployment.