The company that created ChatGPT, OpenAIreported this Friday that it will limit the launch of its new model of artificial intelligence at the request of the president’s administration donald trumpamid an unprecedented process of government review of AI products for cybersecurity risks.
Hours later, its main competitor, Anthropic, announced that the Trump administration authorized a limited launch of its most advanced cybersecurity model, two weeks after the US Department of Commerce effectively banned its use.
Both companies indicated that their new models will only be available to a small group of trusted partners. OpenAI explained that its new AI system, called GPT-5.6 Sol, can only be used by clients approved by the Trump administration.
“We do not believe this type of government access process should become the norm in the long term,” OpenAI said in a statement. The company assured that this testing phase is a temporary measure while it moves towards broader availability in the coming weeks.
The phased rollout of OpenAI’s new system comes after the government took action this month against Anthropic, developer of the Claude chatbot. The company temporarily withdrew two new AI models, Fable 5 and Mythos 5, just days after introducing them, to comply with a Trump directive blocking their use by foreign citizens.
This Friday, the Government partially lifted restrictions on Mythos 5, allowing it to be used again by a small group of cybersecurity experts and critical infrastructure operators, Anthropic reported.
The White House indicated that it continues to collaborate with leading artificial intelligence companies to address the challenges posed by the rapid advancement of this technology.
US authorities have expressed growing concern since Anthropic warned earlier this year that its Mythos model was especially effective at detecting software vulnerabilities that could be exploited by cybercriminals to attack critical computer networks.
Greater oversight over AI models
In early June, Trump signed an executive order that establishes a framework for the federal government to evaluate the national security risks of the most advanced artificial intelligence systems for up to 30 days before their public release. Although companies’ participation is voluntary, the mechanism has not yet been fully implemented.
Some allies of the president have blamed Anthropic and its CEO, Dario Amodei, for driving the need for greater government oversight.
OpenAI stated that its new Sol model (pronounced “sun,” like the word in Spanish) is more effective in helping to detect and correct computer vulnerabilities than in facilitating cyberattacks, and assured that it does not exceed the risk levels established by the company itself. However, he acknowledged that unforeseen risks could arise, especially if the model is combined with other tools.
“The uncertainty, along with the significant leap in its capabilities, is the reason why we are accompanying this launch with greater security measures and gradual distribution,” the company explained.
So far, OpenAI has not revealed the identity of the nearly 20 customers authorized to use the new model.
Criticism of government intervention
Democratic Congresswoman Lori Trahan expressed concern that the Trump administration is deciding which companies can access the most advanced models of artificial intelligence.
“Without a law, without a clear process and without oversight, it is up to designated officials in Washington to decide who has access and who does not,” he said.
Several technology experts also criticized the restrictions placed on Anthropic. Stanford University cybersecurity specialist Alex Stamos maintained that there is no technical basis that justifies the measures taken against the company.
“If the administration really wants the United States to overtake China in this technology race, this is one of the worst decisions it could make,” he said.
Oversight as companies move toward Wall Street
OpenAI CEO Sam Altman spoke this week with US Secretary of Commerce Howard Lutnick as part of a series of negotiations between industry executives and Trump administration officials.
Anthropic has also participated in these conversations, although it maintains a more tense relationship with the Government. The company responded with a lawsuit after the administration ordered federal agencies to stop using its Claude chatbot.
Increased government oversight adds a new challenge for OpenAI and Anthropic, which are exploring the possibility of going public on Wall Street.