The AI standards body under the U.S. National Institute of Standards and Technology (NIST) has experienced another leadership change. Multiple media outlets, citing official confirmation from the agency, reported that Chris Fall, head of the Center for AI Standards and Innovation (CAISI), has departed, just about three months after assuming the role.
This is the second rapid turnover for this position in recent times. The previous appointee, Collin Burns, left after less than a week in office. At the time, The Washington Post reported that Burns was forced out due to his prior employment at Anthropic, amid growing tensions between the Trump administration and the company. The reason for Fall’s departure has not yet been disclosed.
CAISI is responsible for model standards and security testing.
CAISI, under NIST, is primarily responsible for AI model technical standards, testing methods, and cybersecurity risk assessments. In terms of its functions, it should be one of the key agencies in the U.S. federal AI safety framework.
However, CAISI was not included in the list of participating organizations in the new AI safety oversight initiative, "Gold Eagle," launched by the White House this month. The initiative aims to establish a cybersecurity vulnerability coordination mechanism and includes federal agencies such as the U.S. Department of Commerce and the Department of Homeland Security.
The Anthropic incident has intensified regulatory divisions.
The debate over AI regulation in the United States intensified significantly in June. At the time, the U.S. Department of Commerce invoked a rare export control directive, effectively forcing Anthropic to withdraw its Mythos and Fable models from the market. By the end of June, the restrictions were lifted after Commerce Secretary Howard Lutnick endorsed Anthropic’s safety measures.
This incident has also drawn attention to the boundaries of responsibilities among federal agencies. Although CAISI is responsible for standards and testing, recent rounds of practical disputes surrounding the risks of frontier models have not been primarily handled by this agency.
The industry calls for an independent standards organization.
After the Anthropic model resumed operations, Google DeepMind CEO Demis Hassabis publicly called for the establishment of an industry-led, independently operated standard-setting body for advanced AI, modeled after the U.S. Financial Industry Regulatory Authority (FINRA). This proposal overlaps with some of the responsibilities originally assumed by CAISI.
Meanwhile, within the U.S. government, discussions are underway on how to respond to the competitive pressure posed by China’s open-source models. Axios reported that the government previously weighed whether to push for restrictions on Chinese open-source models. Discussions intensified rapidly after the Chinese AI lab Moonshot released a new open-source model, Kimi, last weekend.
Some individuals, including former White House head of AI and crypto affairs David Sacks, have publicly opposed using regulatory tools to protect U.S. closed-source AI companies. Sacks argues that regulation should not become a means of industrial protection.
The testing process still lacks public documentation.
The report noted that CAISI previously released capability reports on Chinese open-source weight models, including Z.ai’s GLM-5.2 and DeepSeek V4 Pro, but still provides limited public details about its own testing procedures. TechCrunch stated that it has repeatedly inquired with the U.S. Department of Commerce and NIST since July 9 about large model evaluation methodologies, but has not yet received a response.
As leadership changes again, the White House launches new initiatives, and calls grow for an independent standards body, the leadership of the U.S. AI safety standards framework remains in flux.