OpenAI is delaying the release of GPT-5.6 at the request of the Trump administration due to safety concerns, limiting access to select clients.
The Trump administration has requested that OpenAI postpone the broad release of its new AI model, GPT-5.6, citing safety concerns. Initially, the model will be accessible only to a select group of clients, as reported by Reuters and The Information.
This request originated from the National Cyber Director's Office and the Office of Science and Technology Policy, aiming to restrict the launch until a framework for assessing AI safety is developed. OpenAI has not publicly confirmed this limited release, but company CEO Sam Altman has reportedly informed staff about it.
Details of the Limited Release
According to The Verge, the limited access will be granted to a small number of corporate clients, with the federal government determining which customers can use the model during its preview phase.
"We have made it clear to the US government that this is not our preferred long-term model, and we will work with them to develop a more sustainable approach to future releases," Altman stated in a message to employees.
Reports indicate that OpenAI has been in discussions with the administration regarding GPT-5.6, and the White House was aware of its capabilities. On June 25, Altman met with US Secretary of Commerce Howard Latnik to ensure government entities could test and approve the model before its release.
Government's Rationale
Media outlets have suggested that the government's intervention is linked to the advanced capabilities of GPT-5.6, which they consider comparable to other high-stakes AI technologies. A White House representative noted ongoing collaboration with AI developers to establish unified approaches for scaling these technologies.
This request aligns with a broader reassessment of US policies on advanced AI. On June 2, an executive order was signed to promote AI innovation while ensuring security, directing various agencies to develop evaluation processes for advanced AI systems. The order also allows companies to share advanced models with the government up to 30 days prior to release, provided they adhere to confidentiality and security measures.
Contrast with Anthropic's Situation
The circumstances surrounding OpenAI differ from the recent case involving Anthropic. On June 9, Anthropic released two versions of its Claude model but faced a government directive on June 12, which restricted access to foreign nationals and was reportedly linked to concerns about security and compliance.
Experts have pointed to the need for clear regulatory frameworks, with Public First CEO Brad Carson emphasizing the importance of fairness in government intervention regarding potentially dangerous AI products.
OpenAI's Regulatory Approach
In early June, OpenAI proposed a governance framework for advanced AI, advocating for enhanced standards and a resilient approach to managing the risks of next-gen models. The company has also filed for a potential IPO, with a valuation target of $1 trillion and plans to go public possibly as early as September, according to reports.
Recently, OpenAI launched GPT-5.5-Cyber, a specialized model for identifying and addressing vulnerabilities, amid the ongoing restrictions faced by Anthropic.
