OpenAI and Anthropic limit latest ChatGPT products to Trump-approved customers amid review

Key Points:

• OpenAI is restricting access to its new AI model, GPT-5.6 Sol, to a small group of government-approved customers following a request from the Trump administration amid concerns over cybersecurity risks.
• Anthropic, OpenAI’s competitor, received limited approval from the administration to release its Mythos 5 model to a select group after previously being banned, while its Fable 5 model remains offline due to government restrictions.
• The Trump administration’s executive order mandates up to 30 days of federal vetting for advanced AI systems to assess national security risks, a process both companies view as temporary but unprecedented in government oversight of AI.
• Critics, including lawmakers and cybersecurity experts, warn that the government’s ad hoc intervention without clear laws or oversight could stifle innovation and harm the U.S. AI industry’s competitiveness against countries like China.
• The increased government scrutiny complicates OpenAI’s and Anthropic’s plans to go public, while Trump has suggested the U.S. government might take equity stakes in leading AI firms to involve the American public as partners.