The US will security check AI models before release

The order was signed by US President Donald Trump on June 2. The ministries of Finance, Defense, Commerce and Homeland Security… are responsible for signing agreements with the unit that owns the AI ​​model, testing it for a maximum of 30 days before they are made public.

In addition, the order also directs US agencies to focus on strengthening cyber defenses across the government. Within 30 days of issuance of the order, the US Department of Defense is required to prioritize cyber defense of information systems by “taking appropriate and expeditious actions as appropriate”.

“This executive order reflects the US administration’s pragmatic approach to collaborating with the AI ​​industry, thereby balancing innovation and security, reinforcing America’s continued global dominance in artificial intelligence and cybersecurity,” a White House spokesperson said.

According to some sources and documents CNN Yes, Mr. Trump planned to pass the order two weeks ago but postponed it just a few hours before signing it. The reason is that the proposal review time is up to 90 days. Meanwhile, the participating business group wants a shorter time in the context of the rapid pace of model development. According to Politicomany large AI companies participated in developing the decree, including Anthropic – a startup that was blacklisted by the Pentagon and considered a “supply chain risk”.

Theo Business InsiderOn May 21, Mr. Trump told reporters at the White House that he postponed the signing due to concerns that it could jeopardize the leading position of American companies compared to China in the AI ​​race.

Chris Lehane, OpenAI’s director of global affairs, called the new move an “important step forward,” emphasizing that safety and innovation must go hand in hand to ensure America’s leadership in AI. “As AI continues to evolve, safe and effective frameworks need to continue to build through democratic institutions, technical expertise and broad input to promote accountability and public trust,” Lehane said.

“The executive order aims to promote innovation while protecting the security of the American people. We welcome this effort,” commented Microsoft President Brad Smith.

Google’s chief legal officer, Kent Walker, also called the order a “negligent step”, while Anthropic affirmed that the company looks forward to cooperating with the White House in supporting enforcement.

Theo Reutersthe executive order signals how the Trump administration is shifting its AI strategy to a more active oversight role. However, this decision could harm the industry’s profits, slowing new product launches or forcing businesses to change the way they operate to address security issues.