- Chris Fall resigned as head of the US AI safety agency CAISI after three months in the role
- Arvind Raman will serve as interim leader of the federal AI testing institute under Commerce
- Fall's predecessor had left within a week
Chris Fall, the head of the US AI safety agency Center for AI Standards and Innovation (CAISI), has stepped down three months after being appointed to lead the federal AI testing institute under the Department of Commerce. Arvind Raman, the Commerce Department official who oversees the institute and is also a former dean of engineering at Purdue University, will take over the role on an interim basis.
The administration has not disclosed why Fall resigned.
Interestingly, the exit comes after another unusually short stint at the agency. His predecessor, Collin Burns, left less than a week after taking office. Reports at the time suggested Burns was forced out because he had previously worked at AI startup Anthropic, which was locked in a dispute with the Trump administration.
In June, the US government temporarily blocked Anthropic's most advanced AI models, Fable 5 and Mythos, citing national security concerns. The restrictions remained in place for nearly three weeks before Commerce Secretary Howard Lutnick lifted the ban, saying the company had addressed the government's safety concerns.
Also read: Mythos A Weapon, Needs Gun License, Testers Warned Anthropic
Fall brought significant government experience to the role. During Trump's first term, he served as director of the Department of Energy's Office of Science and also led the agency's Advanced Research Projects Agency-Energy in an acting capacity. Earlier, he worked at the Office of Naval Research.
CAISI operates under the National Institute of Standards and Technology (NIST) and is responsible for developing technical standards, testing advanced AI models and assessing cybersecurity risks.
However, the agency has recently found itself on the sidelines of some of the administration's biggest AI decisions.
Earlier this month, the White House launched a new AI cybersecurity initiative called Gold Eagle, designed to coordinate vulnerability reporting across federal agencies. While departments including Commerce and Homeland Security were named as participants, CAISI was notably absent.
The leadership change also comes as Washington is debating how to regulate the next generation of AI models.
After Anthropic's models returned to the market, Google DeepMind CEO Demis Hassabis called for an independent, industry-led body to oversee frontier AI, similar to how the Financial Industry Regulatory Authority (FINRA) oversees parts of the financial sector. Building technical standards and evaluating AI systems is already one of CAISI's core responsibilities.
Also read: Superhuman AI Just Years Away, Need Urgent Action Says Google AI Chief
The debate has only intensified following the release of Chinese AI lab Moonshot's latest open-weight Kimi K3 model, which has drawn attention for performing competitively with some of the world's leading AI systems.
Also read: Chinese AI Is Catching Up To The US Faster Than Before: Why It Matters
According to Axios, the Trump administration is now exploring ways to restrict Chinese open AI models, sparking criticism from some within the AI industry. Former White House AI and crypto czar David Sacks argued over the weekend that regulation should focus on security risks rather than be used to shield US AI companies from foreign competition.
CAISI has published technical assessments of Chinese open-weight models, including GLM-5.2 from Z.ai and DeepSeek V4 Pro. But the agency has revealed little about how it evaluates large language models.
