It is reported that the White House has changed its attitude and plans to expand AI supervision, and open models will also be included. According to "Wired" magazine, people familiar with the matter said that White House officials will almost certainly modify the latest AI guidelines formulated by the Trump administration to expand the scope of supervision of AI models.

It is reported that the White House has changed its attitude and plans to expand AI supervision, and open models will also be included. According to "Wired" magazine, people familiar with the matter said that White House officials will almost certainly modify the latest AI guidelines formulated by the Trump administration to expand the scope of supervision of AI models. This month, the White House announced it had developed an AI framework. Under the framework, the most powerful models developed by U.S. AI labs are required to undergo safety testing by the federal government before being released to the public. The U.S. government has not made the framework public and reportedly has no plans to do so. Trump The AI ​​framework currently only works with so-called closed-source models developed by companies such as Anthropic and OpenAI. But a White House official told Wired's Inner Loop column that the framework is expected to include open models in the coming months. In short, White House officials said, once open models reach the same "cutting edge" capabilities as Anthropic's Mythos-level model and OpenAI's GPT-5.6, they will be included in the framework and undergo pre-release testing. The move reflects the White House’s ongoing adjustments to its regulatory approach to the AI ​​industry. The capabilities of AI systems have continued to grow over the past few months, raising concerns that models could autonomously hack into the Pentagon or global financial markets. This concern is not unfounded. OpenAI recently disclosed that during a testing period that lasted several weeks in May and June, a group of AI models worked together on a secret message board to discuss how to gain access to the Internet. After the staff closed the message board, the models re-established the message platform at the end of July and broke through the restrictions again without being discovered. People familiar with the matter said that the White House had hoped to develop a package of guidelines through an executive action, but the exponential development of AI models forced Trump administration officials to update the guidelines in real time. In addition to national security concerns, Trump administration officials are also grappling with the "double standards" that the framework could lead to. If only closed-source models are officially recognized, companies may be wary of open models that don't have equivalent certifications, even if the open models are less expensive. Some Trump administration officials worry that this situation may actually reduce the enthusiasm of American companies to develop open models. However, they also told "Wired" privately that they also realize that enforcing testing requirements that may last up to 30 days may also hinder the development of AI technology. A White House official said the AI ​​framework remains voluntary for now, in large part because U.S. President Donald Trump insists that formal regulation of the industry will only help China catch up with the United States in the AI ​​race. But WIRED has learned that the White House has been facing pressure from other parts of the government