OpenAI's latest artificial intelligence model, GPT-5.6 Sol, will undergo a controlled rollout with a limited preview for trusted partners vetted by the Trump administration. This move represents a significant shift in how advanced AI models are made available, with the U.S. government taking a role in oversight before a broader public release.
GPT-5.6 Sol is described by OpenAI as its most capable model to date, featuring enhancements in coding, biology, and cybersecurity tasks. Initially, access will be restricted to a select group of partners whose participation has been communicated to the government. A wider release through platforms like ChatGPT and the API is planned for a later date, with no general availability announced as of early July.
The administration's interest stems from potential cybersecurity risks associated with the powerful new AI. OpenAI states that Sol is designed to assist in identifying and rectifying software vulnerabilities, rather than executing cyberattacks. However, the company acknowledges that benchmark tests cannot foresee every potential misuse, especially when the AI is integrated with other tools.
This government involvement is part of a broader AI cybersecurity framework initiated by the Trump administration. An executive order from June established a voluntary framework for advanced AI models, suggesting the possibility of secure early access for trusted partners. The administration emphasizes that this framework does not authorize mandatory licensing or pre-clearance for AI model releases, framing the initiative as a measure to address national security concerns.
OpenAI is not the only AI company to face governmental scrutiny over its advanced models. In June, Anthropic was directed by the U.S. government to suspend foreign national access to its Fable 5 and Mythos 5 models, leading to a broader disabling of the models to ensure compliance. While export controls on these Anthropic models were later lifted by the Commerce Department, the episode highlighted ongoing discussions about AI release protocols and the government's role.
These developments raise questions about the balance between leveraging AI for national security and critical infrastructure protection, and the potential for misuse by malicious actors. The decision on who gains early access to such powerful tools, and who is asked to wait, is becoming a central point of debate.
Experts note that AI tools are rapidly advancing, increasing their potential for both beneficial applications and exploitation by scammers. Consumers are advised to exercise caution regarding offers for early access to new AI models, to use strong cybersecurity practices such as antivirus software and password managers, and to be mindful of the information they share with AI tools.