Head of US AI Safety Agency Resigns After Three Months

Chris Fall has resigned as director of the Center for AI Standards and Innovation (CAISI), the federal agency under the US Department of Commerce responsible for safety-testing the latest artificial intelligence models. The departure comes just three months into his tenure and coincides with another key AI policy post in the Trump administration still standing empty.

What happened

The Department of Commerce confirmed Fall's departure to Reuters without giving a reason for the resignation. Fall took charge of CAISI in April, when the Trump administration transformed the former U.S. AI Safety Institute into a new body with a broader mandate combining safety testing with promoting American competitiveness in AI.

Until a permanent successor is named, director duties will fall to Arvind Raman, the current head of the National Institute of Standards and Technology (NIST), which oversees CAISI. Raman previously served as dean of the engineering school at Purdue University.

What CAISI does

The Center for AI Standards and Innovation was created by restructuring the AI safety institute established under the Biden administration. The agency assesses risks tied to the most advanced language models, including their potential use in developing chemical and biological weapons, carrying out cyberattacks, or poisoning training data.

In practice, CAISI works directly with leading AI labs, testing models before they are made publicly available. The agency took on this role after a political dispute over the scope of government oversight of AI development, in which the Trump administration favored lighter regulation than its predecessors.

A vacuum at the top of AI policy

Fall's departure is not an isolated case. The White House adviser post for AI and cryptocurrency, held by David Sacks, has been vacant since March 2026, when his term ended. The lack of a permanent successor means the administration has gone months without a single person coordinating its overall AI technology policy.

The two vacancies in key AI oversight roles have piled up at a moment when the United States is competing with China for technological dominance in AI, while Congress and the White House still have not agreed on a coherent approach to regulating the sector.

Why it matters for the industry

For companies developing AI models, including OpenAI, Google DeepMind, Anthropic, Microsoft and xAI, CAISI is one of the few points of contact with the US government for voluntary safety testing before new systems launch. The lack of stable leadership at the agency raises questions about the continuity of those relationships and the pace of risk assessments for the latest models, the kind that have recently been hitting the market at breakneck speed.

For Polish companies and institutions tracking global AI safety standards, the situation in Washington matters indirectly: guidelines from agencies like CAISI often become a reference point when building risk-assessment frameworks of their own, alongside the rollout of the EU's AI Act.

The Department of Commerce has not given a timeline for naming Fall's permanent successor. Until then, the agenda for model testing and risk assessment remains in the hands of NIST's interim leadership.