The Trump administration has tightened controls on the release of AI models, weakening the voice of technology giants. On July 18, according to CNBC, two people familiar with the matter said that the Trump administration has taken new measures to strengthen control over the future AI model release process by stipulating which companies and entities are allowed to access the latest cutting-edge models.

The Trump administration has tightened controls on the release of AI models, weakening the voice of technology giants. On July 18, according to CNBC, two people familiar with the matter said that the Trump administration has taken new measures to strengthen control over the future AI model release process by stipulating which companies and entities are allowed to access the latest cutting-edge models. Prior to this, this decision-making power was in the hands of American AI giants. Anthropic and OpenAI had previously decided on their own which businesses and institutions could use their most powerful models, often including large enterprise customers. Trump However, the U.S. government has recently begun to tighten regulations on the release of AI models. Anthropic has released its most capable Mythos cybersecurity model to only a handful of partners through its Project Glasswing. OpenAI has restricted access to its recently released GPT-5.6 at the request of the government, and its network security model also has a similar cooperation alliance called "Daybreak". A White House official told CNBC that the government will not approve the release of AI models from private companies. The official said any contact, testing or meetings with government experts would be "voluntary" and that "decisions on the timing and scope of releases are entirely at the company's discretion," referring CNBC to Trump's recently signed executive order. "My administration continues to work with all cutting-edge AI labs in the United States to strengthen the safety of this technology without stifling innovation," the official said. However, last month, the Trump administration blocked access to the Claude Mythos 5 and Fable 5 due to "national security concerns." The government restored access after weeks of intense negotiations with Anthropic. OpenAI said last month that it would only make new AI models available to "trusted partners" in order to comply with government requirements. CNBC pointed out that the White House is currently walking a tightrope on the issue of AI regulation. On the one hand, advanced AI tools bring huge cybersecurity risks; on the other hand, China’s cheaper open weight models are rapidly narrowing the gap with the United States’ cutting-edge AI laboratories. via cnBeta.COM - Chinese industry information website (author: Source: Phoenix Technology)