Rebooting Democracy in the Age of AI

Building domestic AI systems or reducing reliance on big tech is not the same as making AI work for the public. Examples such as Spain’s ALIA project, the EU AI Act, and Italy’s new AI law show that ownership and regulation go only so far. What really matters is whether public institutions have the people, skills, and authority to oversee these systems once they’re in use.

A new paper urges universities to resist the uncritical adoption of AI, arguing that existing research integrity standards already prohibit much of what institutions are normalizing. Beth Simone Noveck's diagnosis is sharp: vendor dependency, opacity, and hype are reshaping public institutions from the inside. But for governments, refusal isn’t an option. The real question is not whether to adopt AI, but who controls it and on what terms.

As governments move from AI pilots to workforce-wide access, the hardest questions are practical: What does it cost at scale? Who gets access? How do you log usage and manage risk? How do we pair access with training so it’s safe — and actually useful?Drawing on lessons from New Jersey and Boston, this post outlines how to structure AI deployment using usage-based pricing, cloud infrastructure you control, and training designed to drive real impact.