The Constitutionality of AI-Driven Grant Cancellations

The intersection of AI automation and government administration is often framed as a growth opportunity for efficiency. However, a recent federal court ruling has highlighted the critical risks associated with delegating high-stakes administrative decisions to AI systems. A federal judge has ruled that the Department of Government Efficiency (DOGE)'s cancellation of humanities grants was unconstitutional, citing a specific failure in the process used to identify and terminate funding.

The Core of the Legal Challenge

At the heart of the legal battle was the attempt to use AI to purge grants associated with Diversity, Equity, and Inclusion (DEI) initiatives. The government's defense rested on the a claim that the government itself was not engaging in viewpoint discrimination, but rather that any such classification was the result of the AI's internal logic—specifically, ChatGPT.

Judge McMahon, however, rejected this argument, establishing a a key legal precedent regarding the agency of AI in government operations. The court found that because the government chose to employ AI as its instrument, it remains legally responsible for the outcomes of that instrument's output.

"ChatGPT was the Government’s chosen instrument for purposes of this project, and DOGE’s use of AI to identify DEI-related material neither excuses presumptively unconstitutional conduct nor gives the Government carte blanche to engage in it," wrote Judge McMahon.

Administrative Failure and Lack of Expertise

Court documents reveal a significant gap in professional expertise among the personnel overseeing the process. The ruling detailed that the second stage of the grant termination process involved members of DOGE’s “Small Agencies Team,” including Justin Fox and Nate Cavanaugh.

According to the court record, neither Fox nor Cavanaugh had prior experience in government, public or private grant administration, or the scholarly review of humanities projects. The court noted that as both individuals were in their twenties, they lacked the experience necessary to evaluate the merit of the humanities projects they were overseeing the cancellation of.

Implications for AI Governance

This case serves as a critical case study for the rest of the public sector. It demonstrates that the 'black box' nature of AI can no longer be used as a shield for administrative actions that violate constitutional protections.

For technical leaders and policy makers, the takeaway is clear: AI can be used to assist in the process, but it cannot replace the human oversight and subject-matter expertise required for legal and constitutional compliance. The delegation of high-stakes decisions to an untrained workforce using an unverified AI tool leads not only to efficiency gains, but to significant legal liability.