US AI Safety Chief Chris Fall Resigns After 3 Months
- Center for AI Standards and Innovation (CAISI) three months after his appointment, according to a CNBC report published July 20, 2026.
- Arvind Raman, the director of the National Institute of Standards and Technology (NIST), will serve as the acting director of CAISI while maintaining his current oversight of NIST,...
- Fall's exit leaves the White House without a designated pointperson for AI research and policy.
Chris Fall has resigned as the head of the U.S. Center for AI Standards and Innovation (CAISI) three months after his appointment, according to a CNBC report published July 20, 2026. A Commerce Department spokesperson confirmed the departure without providing a specific reason for the resignation.
Arvind Raman, the director of the National Institute of Standards and Technology (NIST), will serve as the acting director of CAISI while maintaining his current oversight of NIST, CNBC reported citing the spokesperson.
Leadership Vacancies in U.S. AI Policy
Fall’s exit leaves the White House without a designated pointperson for AI research and policy. This follows the departure of David Sacks, the former White House AI and crypto czar, who stepped down from his role in March 2026 and has not been replaced, according to CNBC.
The position of CAISI director has seen significant instability. The Hill reported on July 20, 2026, that Fall’s predecessor, Collin Burns, was removed by the White House four days after starting the role. The report stated Burns was pushed out due to ties with Anthropic that clashed with the administration.
CNBC further reported that Fall’s tenure in the position was always intended to be temporary.
CAISI Mandate and Private Sector Agreements
CAISI is a division of the Commerce Department tasked with managing risks associated with advanced AI models. According to Reuters, the agency’s responsibilities include working with leading AI labs to test unreleased models for vulnerabilities, calculating the risks posed by these models, and limiting the ability of U.S. adversaries to utilize AI.
The agency maintains formal agreements with several major AI developers to facilitate national security testing. PYMNTS reported in May 2026 that Google DeepMind, Microsoft, and xAI signed agreements to share their frontier AI models with CAISI before public release.
These companies joined Anthropic and OpenAI, which signed agreements with the predecessor to CAISI in August 2024. Those two firms later renegotiated their terms to align with CAISI’s specific directives, according to the PYMNTS report.
Executive Oversight and Cybersecurity
The agency’s operations align with a June 2026 executive order signed by Donald Trump. That order requires companies to provide the White House with access to new AI models prior to release and participate in benchmarking to evaluate model capabilities. The order also mandates that trusted government partners be granted access to models to improve the cybersecurity of critical infrastructure.
.jpg)
Despite these mandates, CAISI’s role in new initiatives appears limited. CNBC reported that when the White House announced Gold Eagle, a clearinghouse designed to identify and fix cyber vulnerabilities, CAISI was not listed as an agency involved in the development of the program.
