US AI standards chief resigns three months after appointment

Chris Fall has resigned as director of the Center for AI Standards and Innovation three months after the Trump administration selected him for the post, CNBC reported Monday. The departure places interim leadership at a Commerce Department unit that is meant to support federal testing and research on commercial artificial intelligence systems.

Kristen Eichamer, a Commerce Department spokesperson, told CNBC that Arvind Raman, director of the National Institute of Standards and Technology, will serve as acting director of CAISI and continue to oversee the agency. Eichamer did not give CNBC a reason for Fall’s resignation.

CAISI sits within the U.S. Department of Commerce. According to the center’s website, its role is to help the government conduct “testing and collaborative research” involving commercial AI systems. In practice, that remit places the center near a central policy issue for Washington: how federal agencies assess increasingly capable AI models before or around their release to the market.

Leadership change amid AI policy buildout

Fall’s exit comes as the Trump administration is building a more active federal process for AI oversight. President Donald Trump signed an AI executive order in June that asks AI developers to provide models to the government on a voluntary basis so their capabilities can be assessed before full release, according to CNBC. The order also gave federal agencies 60 days to produce an evaluation framework.

The leadership picture around federal AI policy has already shifted this year. Venture capitalist David Sacks, who had served as the White House AI and crypto czar, stepped down from that role in March, CNBC reported. The role has not been filled, according to CNBC.

The administration’s policy process is developing while Chinese open-source models are gaining attention in the United States. CNBC reported that Chinese startup Moonshot AI last week introduced Kimi K3, a model the company says narrows the gap with leading U.S. systems and outperforms the most capable models from OpenAI and Anthropic on some benchmarks.

Model releases have faced federal constraints

The federal review process has already affected some model launches, according to CNBC. OpenAI said in June that, at the U.S. government’s request, it limited release of its GPT-5.6 model series to a set of “trusted partners.” Anthropic, two weeks earlier, disabled access to its Fable 5 and Mythos 5 models to comply with a Commerce Department export-control directive, CNBC reported.

Both companies later released their models more broadly, according to CNBC. The episodes illustrate how voluntary assessment, export controls and model access decisions can overlap when companies seek to launch frontier systems.

The White House has also announced a clearinghouse known as “Gold Eagle” as part of implementing the AI executive order. According to a White House release cited by CNBC, Gold Eagle is designed to identify and address cybersecurity vulnerabilities, “intake and prioritize identified cybersecurity vulnerabilities” and “coordinate scanning verifications.” CNBC has reported that the clearinghouse gives the White House authority to approve which companies can access advanced AI models.

CAISI was not listed by the White House as one of the agencies involved in developing Gold Eagle, according to CNBC. With Raman now serving as acting director, the Commerce Department’s AI standards unit remains part of a wider federal effort to define how the U.S. government tests and monitors advanced commercial AI systems.

This story draws on original reporting from CNBC.