Chris Fall has resigned as director of the Center for AI Standards and Innovation, the federal body responsible for testing frontier AI models, just three months after being appointed to the role, a Commerce Department spokesperson confirmed Monday.
Arvind Raman, director of the National Institute of Standards and Technology and a former dean of engineering at Purdue University, will serve as acting director of CAISI while the administration searches for a permanent replacement, according to the Commerce Department. Spokesperson Kristen Eichamer did not disclose a reason for Fall’s departure, and a spokesperson for President Trump declined to comment. Fall could not immediately be reached.
CAISI, housed within the Commerce Department, is responsible for developing federal testing and evaluation capabilities for advanced AI systems and works with major AI developers, including Anthropic, Google DeepMind, OpenAI, Microsoft, and xAI, to test unreleased models for security risks. The agency is the direct successor to the U.S. AI Safety Institute, established under the Biden administration, which the Trump administration reorganized and rebranded after taking office.
Fall was appointed to lead CAISI in late April, following an earlier administrative shuffle. The administration had initially sought Collin Burns, a former researcher at OpenAI and Anthropic, for the role, but the Commerce Department pivoted to Fall while Burns was still completing onboarding, according to AndroidHeadlines. Fall previously directed the Office of Science at the Department of Energy during Trump’s first term.
Fall’s exit is the latest leadership change at the agency in little more than a year. Elizabeth Kelly, the AI Safety Institute’s inaugural director under the Biden administration, departed in early 2025 shortly after Trump returned to office and rescinded the executive order that had created the institute.
The resignation lands weeks after CAISI’s testing process featured directly in two high-profile disputes over frontier AI releases. Anthropic disabled access to its Claude Fable 5 and Mythos 5 models last month to comply with a government export control directive, later restoring access after revising its safety classifiers. OpenAI separately said it agreed to limit its GPT-5.6 model series to a group of “trusted partners” at the government’s request. Both episodes followed President Trump’s June executive order asking AI developers to voluntarily provide the government early access to frontier models for evaluation, and directing federal agencies to develop a formal testing framework within 60 days.
Fall’s departure adds to broader uncertainty over who is directing AI policy inside the administration, following CNBC’s reporting last week that the White House has moved to require government approval over which companies and entities can access partner programs run by Anthropic and OpenAI, arrangements the labs previously controlled themselves.
CAISI has not indicated when a permanent director will be named.
