Americans Don't Trust Government With Their Data — And They Have Good Reason Not To

NewsOnScale Staff

September 26, 2026

There is a version of the AI-in-government story that gets told frequently in policy circles and technology press releases: the story of efficiency, of streamlined services, of backlogs cleared and forms processed at machine speed. What that story tends to leave out is the person on the other end — the citizen whose medical history, financial records, immigration status, or criminal record sits inside a database they cannot audit, correct, or meaningfully consent to sharing.

New survey data from the Center for Democracy and Technology puts that person back at the center of the conversation. The findings show that Americans across the political spectrum are worried about personal data held by public agencies, and that significant majorities want government to be held to higher accountability standards when it comes to how that information is collected and used.

This is not a fringe concern. It is a democratic one.

## What the Data Actually Shows

The CDT survey captures something that is easy to miss when the policy conversation focuses on capability: public trust in government data stewardship is eroding, not building. Americans are not simply anxious about private tech companies — they are anxious about the state itself acting as a data broker, an inference engine, and a surveillance apparatus.

That anxiety is grounded in documented reality. Federal and state agencies have for years shared data with law enforcement, immigration authorities, and third-party contractors under legal frameworks that most citizens would not recognize if explained to them. The Privacy Act of 1974 — the primary federal statute governing how agencies handle personal records — was written before the internet existed. It has not been comprehensively updated since.

Into this already strained accountability environment, AI is now being introduced at scale. Agencies are using algorithmic tools to assess benefits eligibility, flag fraud, prioritize audits, and screen job applicants for public sector positions. Most of these deployments involve no mandatory disclosure, no independent audit requirement, and no meaningful recourse for individuals whose outcomes are shaped by systems they cannot see.

## The Gap Between Transparency and Accountability

This is where the CDT findings intersect with a broader structural problem in civic technology. Transparency — publishing a dataset, releasing an algorithm's name, posting a privacy policy — is increasingly understood by researchers and practitioners alike as insufficient on its own. What the public is asking for, even if not always in these precise terms, is accountability: the ability to contest decisions, to understand how data was used, and to impose consequences when agencies get it wrong.

The distinction matters enormously for AI systems in particular. A machine learning model that produces discriminatory outputs can be disclosed and still cause harm. An agency can publish its data retention policy and still share records in ways that damage lives. Transparency creates the conditions for accountability — it does not deliver accountability by itself.

Government AI deployments that are marketed as transparent while remaining structurally unaccountable represent a specific kind of legitimacy laundering that citizens, advocates, and journalists should be equipped to identify.

## Why This Belongs in the AI Agent Economy Conversation

The emergence of AI agents — systems that don't just analyze data but take actions on behalf of users or institutions — raises the stakes considerably. When an AI agent operates inside a public agency, making or influencing decisions about benefits, enforcement, or access to services, the data it draws on and the decisions it produces become civic infrastructure. The accountability frameworks governing that infrastructure are, at present, dramatically underdeveloped relative to the technology being deployed.

The CDT survey suggests the public understands this intuitively, even without the technical vocabulary. They know something important is happening to their data. They know accountability mechanisms are weak. And they want that to change.

Building those mechanisms — through legislation, procurement requirements, independent audits, and enforceable individual rights — is not a technical problem. It is a political one. The question is whether elected officials will treat public anxiety about government data as a mandate for reform, or simply as another number in a poll.