President Donald Trump signed an executive order on Tuesday that seeks more oversight over new artificial intelligence models before companies release them to the public. After months of debate over how to handle AI, Trump took his biggest step yet in regulating the quickly developing tech. Trump’s executive order asks AI companies to voluntarily submit their most powerful models to the government to test for up to 30 days before releasing the models to the public or granting access to other partners. “Advanced AI capabilities make our Nation stronger, but also introduce new national security considerations that require coordinated action across executive departments and agencies,” Trump’s order stated. It added that the Trump administration would work to ensure that “the best and most secure technology is deployed rapidly to confront any and all threats to our country.” WATCH FROM TODAY’S EPISODE OF MORNING WIRE:    The move is a shift in Trump’s previous strategy, which took a more hands-off approach in his goal to place American tech companies ahead of China and bolster the economy. The order also asks the Treasury secretary to create an AI “cybersecurity clearinghouse” designed to review security vulnerabilities AI discovers.  Trump signed the executive order as security concerns rise. Anthropic recently said that its most powerful AI model, Mythos, was too dangerous to be released to the public. The company said Mythos “poses unprecedented cybersecurity risks,” and the company shared its concerns with the government. Anthropic added that the model is “too effective at finding high-severity vulnerabilities, or potential weaknesses, in major operating systems and web browsers,” a danger that can be exploited by cybercriminals and spies.  Last month, Trump nearly signed an executive order that would review new AI models for up to 90 days, but he postponed that signing just hours before it was set to take place. Trump told reporters he delayed the original signing because he thought the executive order “gets in the way” of AI development and he “didn’t like certain aspects” of it.