Chris Fall, director of the Center for AI Standards and Innovation, resigned on Monday just three months after being appointed to lead the federal AI testing institute housed within the Commerce Department.
The Commerce Department confirmed Chris Fall's departure and said a new director will be announced in the coming weeks
J. David Ake / Getty Images
Chris Fall, director of the Center for AI Standards and Innovation, resigned on Monday just three months after being appointed to lead the federal AI testing institute housed within the Commerce Department.
A Commerce Department spokesperson confirmed the resignation. No reason for Fall's departure was provided, and a spokesperson for President Donald Trump declined to comment, according to Reuters.
Arvind Raman, director of the National Institute of Standards and Technology, has been designated acting CAISI director as the administration looks for someone to fill the role on a permanent basis. Raman joined NIST as its director on June 30, having previously served as dean of engineering at Purdue University. The Commerce Department expects to announce a new director in the coming weeks, according to Axios.
According to Axios, a Commerce official indicated that Fall was never meant to stay in the role long-term and that Raman has spent the past several weeks evaluating potential candidates.
Fall took charge of CAISI in April, following the Trump administration's restructuring of what had been called the U.S. AI Safety Institute. CAISI's core mission involves building out AI testing and evaluation tools while helping to shape standards governing advanced AI systems. It works with leading AI labs — including Anthropic, Google $GOOGL's DeepMind, OpenAI, Microsoft $MSFT, and Elon Musk's xAI — to test unreleased models for vulnerabilities, with a focus on limiting risks that U.S. adversaries could exploit to develop chemical or biological weapons or corrupt AI training data.
The departure is the latest disruption for an agency that has faced repeated changes in direction. CAISI removed from its website the details of agreements under which Google, Microsoft, and xAI had committed to submitting new AI models to government scientists for security testing before public release. The removal drew attention in May, with no public explanation offered. The agreements had been announced on May 5 under a framework that extended arrangements CAISI had already established with OpenAI and Anthropic.
The Trump administration has also been navigating broader questions about AI oversight. A program called Gold Eagle would give the federal government authority to determine which partners can access the most capable AI models from Anthropic and OpenAI, according to CNBC — a shift that goes beyond the voluntary framework President Trump established in a June executive order asking companies to submit new models for government review before release.
Join 500,000+ readers who start their day with Quartz.
By subscribing, you agree to our Terms of Service and Privacy Policy.