OpenAI, the creator of ChatGPT, announced on Friday that it is limiting the launch of its latest artificial intelligence model at the request of the U.S. President Donald Trump’s administration. This move is part of the government’s efforts to assess potential cybersecurity risks associated with AI products.
The new AI model, named GPT-5.6 Sol, will initially only be accessible to a select group of approved partners designated by the Trump administration. OpenAI emphasized that this government vetting process should not become a permanent practice and sees this restricted release as a temporary measure before wider availability in the near future.
In a similar vein, Anthropic, a competitor of OpenAI known for the Claude chatbot, recently withdrew its latest AI models, Fable 5 and Mythos 5, shortly after their public release to comply with a directive barring their use by foreign individuals.
The White House confirmed its ongoing collaboration with leading AI research labs to address the challenges of rapidly advancing AI technology. Concerns have been mounting following Anthropic’s caution earlier this year about the potential weaponization of its Mythos model by malicious actors to exploit software vulnerabilities and jeopardize global computer networks.
This action by OpenAI aligns with an executive order signed by President Trump in June, establishing a framework for evaluating the national security implications of advanced AI systems before their public rollout. OpenAI stated that its Sol model prioritizes identifying and remedying vulnerabilities over engaging in cyberattacks and does not surpass the company’s risk threshold. However, recognizing potential unforeseen risks, especially when combined with other tools, the company is implementing enhanced safeguards and a phased release strategy for the model.
Criticism has arisen from a wide array of cybersecurity experts over the government’s handling of Anthropic’s models, particularly the shutdown of Fable. Stanford University cybersecurity expert Alex Stamos disputed the factual basis for these actions, underscoring that the cybersecurity community perceives no unique risks associated with these models compared to other publicly available AI technologies, including those originating from China. Stamos, who serves as the chief product officer at AI security firm Corridor, emphasized the detrimental impact of stifling innovation in the U.S. in the global AI race.
