Under an executive order issued by President Donald Trump, AI companies such as OpenAI and Anthropic are being asked to submit advanced artificial intelligence models to the government for vetting on cybersecurity, confidentiality, "insider risk" and intellectual property protections.
But as the executive order makes clear, AI companies are under no obligation to do much of anything.
At the end of a section ordering the creation of a vetting process in which AI companies would submit their models 30 days ahead of release, it states: "Nothing in this section shall be construed to authorize the creation of a mandatory governmental licensing, preclearance, or permitting requirement for the development, publication, release, or distribution of new AI models, including frontier models."
A White House representative did not immediately respond to a request for comment.
Frontier models are cutting-edge AI models that could pose a major security risk, such as Anthropic's Mythos, which the company kept from public release due to cybersecurity concerns. US-based AI companies are in a race to train and release artificial intelligence models, especially amid competition from Chinese companies.
But the rapid release cycle of new models has raised concerns that they aren't being properly tested or regulated before going public, particularly regarding their human impacts or potential use as hacking tools.
The executive order also includes provisions for agencies such as the Pentagon and the US Department of the Treasury to beef up their cybersecurity defenses over the next 30 days. Over the next 60 days, agencies, including the Department of Homeland Security and the National Security Agency, are expected to create a framework for evaluating AI models, even though companies aren't required to submit them.
(Disclosure: Ziff Davis, CNET's parent company, in 2025 filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.)
The 'appearance' of oversight?
Executive actions similar to this one were expected two weeks before the June 2 order, but the final version included several notable changes. According to CNN, the original draft called for a 90-day review period instead of 30 days. However, AI companies involved in shaping the order, including Anthropic, reportedly pushed back on the longer timeline.
CNN also reported that a Department of Commerce Institute of Standards and Technology announcement last month related to AI companies sharing their AI models with the government has disappeared from the agency's website.
Given the lack of enforcement provisions or mandatory requirements for AI companies, the executive order received a muted reaction.
"Voluntary frameworks are not enough," Anthony Aguirre, CEO and president of the AI safety nonprofit Future of Life Institute, said in a statement emailed to CNET. "We need a mandatory government pre-deployment review process for the most powerful AI systems, allowing the government to block the release of systems that pose an unacceptable national security risk."
The executive order doesn't make clear what happens if the government receives early access to an AI model and a significant problem is discovered.
John Thickstun, assistant professor of computer science at Cornell University, told CNET that this lack of clarity makes it hard to understand what the executive order will achieve.
"Without clearer answers to these questions, my read of this is that it creates some appearance of oversight while largely continuing the administration's hands-off approach to AI governance," Thickstun said.
