Summary
- The Trump administration has reportedly requested that OpenAI restrict the initial launch of GPT-5.6 to select government-approved partners.
- Officials are assessing the model under a new federal framework for advanced AI testing.
- This action follows years of demands from AI developers for enhanced oversight of cutting-edge models.
Reports indicate that President Donald Trump’s administration has asked OpenAI to confine the initial deployment of GPT-5.6 to a limited number of government-sanctioned partners while federal representatives review the model. This information was shared by The Information and Axios.
This request represents the second instance this month in which the U.S. government has intervened to restrict the release of an advanced AI model, following its directive to Anthropic to halt public access to Claude Fable 5 and Mythos 5 due to national security concerns.
The White House's Office of the National Cyber Director and the Office of Science and Technology Policy reportedly urged OpenAI to limit the rollout of GPT-5.6 as the administration works on establishing a framework for evaluating advanced AI models prior to broader distribution. Sources familiar with the matter have indicated that the request was influenced by GPT-5.6's "Mythos-like" functionalities rather than indicating a wider change in AI policy.
This request follows President Trump's executive order issued earlier this month, which tasked federal agencies with developing a voluntary testing framework for advanced AI systems before their release, after extensive internal discussions regarding the program's structure.
This action also highlights a changing dynamic between major AI developers and the federal government, particularly after years of calls from the development community for the establishment of industry regulations.
In 2023, during Senate testimony, OpenAI CEO Sam Altman advocated for the creation of a regulatory body to oversee advanced AI systems, asserting that independent oversight will eventually be essential. Recently, Anthropic CEO Dario Amodei contended that the most capable AI models ought to undergo thorough government-backed assessments prior to deployment due to their potential to facilitate advanced cyberattacks, biological weapons development, and other national security risks.
These discussions have become increasingly formalized, with Anthropic, OpenAI, and Google each have put forward proposals detailing governance for frontier AI. Although their methods vary, all three advocate for structured evaluations of high-capability models, increased transparency in safety assessments, independent reviews of high-risk systems, and a more significant governmental role in overseeing AI development.
The administration's intervention may also serve as a test of whether the governance frameworks proposed by leading AI firms can be implemented consistently across the industry. Critics caution that if the largest AI companies influence regulations that are then applied unevenly, it could lead to a situation of regulatory capture, thereby benefiting a select group of companies and hindering competition.