US AI safety center may report to White House office

A shift in reporting lines

The US government’s AI safety center, known as the Center for Advancing Innovation and Standards for Super Intelligence (CAISSI), may be moving closer to the executive branch. According to a report by the Associated Press, officials at the center were recently asked to informally report to an office of the White House.

This information comes from two former government officials who spoke on condition of anonymity. While the specific White House office has not been named, the change suggests a shift in how the Commerce Department manages its AI safety oversight. The report highlights a move toward more direct involvement from the presidency in managing super intelligence risks.

The center was recently renamed from the U.S. Center for AI Standards and Innovation. This transition occurs as the government grapples with how to balance innovation with safety. The lack of a written directive or formal order makes the exact nature of this reporting line difficult to confirm at this stage.

Concerns over transparency

The potential move to a White House reporting structure has sparked concern among critics regarding the independence of the center. Some worry that the executive branch might politicize decisions regarding serious risks, such as cybersecurity threats or bioweapons. There is a fear that findings could be altered or muted to show deference to the tech industry.

Amos Toh of the Brennan Center told the Associated Press that the decisions regarding these risks "should be explained and rationalized in the public eye." This sentiment reflects a broader concern that moving oversight into the White House could decrease public accountability. Critics argue that transparency is essential for maintaining trust in government safety evaluations.

Conversely, some observers suggest that a closer relationship with the President could be beneficial for the center's growth. They argue that this proximity might give the division more prominence and potentially more funding. This creates a tension between the need for independent oversight and the desire for high-level political support.

Current status and leadership

CAISSI currently faces significant internal leadership challenges and a relatively small budget. The center is allocated $10 million a year, which is roughly one-eighth of the annual budget of its UK counterpart. The division has struggled to maintain stable leadership, with a director being hired and fired within days in April, followed by a resignation from the next replacement.

The head of the parent institute currently serves as the acting director for the center. This leadership instability has coincided with a period of reduced public output. Congress began asking in August why the center had stopped making its reports on AI system safety public, a move that has drawn attention from policymakers.

Despite these hurdles, the center remains a key point of contact for the industry. It serves as the primary hub for testing and research on commercial super intelligence systems. It is tasked with establishing voluntary agreements with private sector developers and evaluators to ensure safety standards are met.

Testing and industry impact

The center plays a vital role in coordinating evaluations with major technology companies. NIST has announced agreements with Google DeepMind, Microsoft, and xAI for pre-deployment evaluations. These build on existing agreements established in 2024 with OpenAI and Anthropic, totaling more than 40 evaluations to date.

While these five labs have testing agreements, the center's recent public model assessments have focused on other entities. For example, NIST’s pages list assessments for DeepSeek V4 Pro, Z.ai’s GLM-5.2, and a joint assessment of Kimi K3’s cyber capabilities. These specific models were not part of the primary testing agreements with the five major labs.

The center also works with other government agencies to develop incident-response practices. In October, it published a playbook alongside the NSA’s Super Intelligence Security Center and CISA. These collaborations show that while the reporting line is changing, the center's technical role in security and evaluation remains a core function.

The road ahead for safety

The White House has not provided a specific response to questions regarding CAISSI's status. Instead, it issued a statement regarding the establishment of a new Super Intelligence Force. This new force, chaired by Jay Clayton, is intended to ensure continued U.S. leadership on super intelligence for the benefit of the American people.

The shift in reporting lines reflects a broader trend of integrating AI safety into national security and geopolitical alignment. As the government moves to define