OpenAI agreed to stagger the rollout of GPT-5.6 Sol, directing early access only to “Trump-approved” government-cleared customers and limiting deployment to a small preview group. The move follows security concerns in Washington about advanced cyber capabilities seen in frontier models. The company framed the step as part of ongoing engagement with the U.S. government, while saying it does not want the process to become a “long-term default.” The change also reflects a pattern already visible this month in Anthropic’s restricted access after Commerce actions. For colleges and research institutions looking to use frontier AI tools, the practical impact is access variability—who can deploy the newest capabilities, when, and under what documentation—especially for cybersecurity and sensitive academic workflows.