OpenAI Restricts New AI Model Release at Trump Administration's Request
The tech company is limiting access to its GPT-5.6 Sol model during a government cybersecurity review, a move that follows similar actions against competitor Anthropic.
OpenAI, the creator of ChatGPT, announced Friday that it is restricting the release of its latest artificial intelligence model, GPT-5.6 Sol, at the request of the Trump administration. This decision marks a significant instance of government vetting for AI products due to potential cybersecurity risks.
GPT-5.6 Sol will initially be available only to a select group of "trusted partners" who have received approval from the administration. OpenAI stated that it does not view this type of government access process as a suitable long-term default, describing the current testing period as a temporary measure towards broader availability in the coming weeks.
The phased release of OpenAI's new AI system comes after similar actions taken against its rival, Anthropic, the developer of the Claude chatbot. Earlier this month, Anthropic temporarily removed two new AI models, Fable 5 and Mythos 5, from public access following a directive from the Trump administration that blocked their use by foreign nationals.
The White House indicated it is continuing to collaborate with leading AI laboratories to address the challenges associated with scaling rapidly advancing AI technologies. Concerns have heightened following Anthropic's earlier warning that its Mythos model demonstrated an ability to identify software vulnerabilities in ways that could be exploited by malicious actors, potentially jeopardizing critical computer networks globally.
In June, President Trump signed an executive order establishing a framework for federal agencies to review the national security risks of advanced AI systems for up to 30 days before their public release. While participation from AI developers is described as voluntary, the framework is still under development.
OpenAI asserts that its new Sol model is more proficient at assisting users in identifying and rectifying vulnerabilities than at executing cyberattacks and does not breach the company's internal risk thresholds. However, the company acknowledged potential unforeseen risks, particularly if the model is integrated with other tools. "That uncertainty, along with the model's broader step change in capabilities, is why we are pairing the model's increased capabilities with stronger safeguards and a phased release," OpenAI stated.
A segment of cybersecurity experts has voiced criticism regarding the government's interventions, which led Anthropic to disable Fable, a model marketed as a more secure alternative to Mythos. Alex Stamos, a cybersecurity expert from Stanford University and chief product officer at Corridor, stated that the cybersecurity industry largely questions the factual basis for these actions. He noted that his review of an analysis of Fable's research did not reveal risks distinct from those present in other publicly accessible AI models.
OpenAI CEO Sam Altman reportedly discussed the model's release with U.S. Commerce Secretary Howard Lutnick on Wednesday, as part of ongoing negotiations between AI industry executives and administration officials. Anthropic has also participated in these discussions, although its CEO Dario Amodei has had a more challenging relationship with the administration. The Pentagon previously identified Anthropic as a national security risk due to concerns raised about AI in warfare, and President Trump had ordered federal agencies to cease using Claude, prompting Anthropic to file a lawsuit.