The independence of the U.S. government’s primary AI safety evaluation center is facing scrutiny after two former government officials told the Associated Press (AP) that staff had been asked to informally report to a White House office. The White House did not respond to the publication’s specific questions about the arrangement.

The arrangement has raised concerns that political involvement could influence independent assessments of advanced AI systems, according to an AP investigation published November 8.

The Center for Advancing Innovation and Standards for Super Intelligence (CAISSI), housed within the Commerce Department, evaluates the safety and security of advanced AI systems. It currently operates without a permanent director and has stopped publicly releasing some research.

The development comes as the Trump administration favors voluntary industry commitments for AI safety. This raises questions about who independently verifies companies’ safety claims, particularly when technology companies can restrict evaluators’ access to their models.

Implications for AI Governance

For chief data officers (CDOs), the debate highlights the importance of independent AI evaluation, transparent testing and accountability, which are all critical to AI governance. 

As organizations deploy AI agents capable of executing multi-step tasks, assessments must examine their actions and decisions throughout a process, not just their final outputs.

Without credible independent oversight, organizations may have to rely heavily on vendors’ own assurances about the safety and reliability of their systems.