When Sunlight Isn't Enough: Why Open Data Alone Won't Fix Broken Government

NewsOnScale Staff

July 18, 2026

There is a seductive logic to the transparency agenda. Post the data online. Publish the contracts. Open the budget spreadsheets to the public. The theory holds that an informed citizenry will do the rest — that exposure is, in itself, a corrective force.

It isn't. And the civic technology community is finally saying so out loud.

## The Disclosure Paradox

Research from Harvard's Ash Center for Democratic Governance and Innovation cuts to the core of a tension that practitioners in government accountability work have quietly acknowledged for years: transparency is a necessary condition for fighting corruption, but it is nowhere near a sufficient one. Publishing data does not automatically produce the capacity to interpret it, the media infrastructure to amplify it, or the institutional will to act on it.

This matters enormously for how we think about AI-powered civic tools — one of the fastest-growing categories of investment in the government technology sector right now. If the foundational assumption underlying those tools is that more information flow equals better governance, then billions of dollars in platform development and grant funding may be optimizing for the wrong outcome.

The Ash Center's framing draws on decades of comparative anticorruption research across multiple countries and political systems. What that body of evidence consistently shows is that transparency interventions work best when they are embedded in ecosystems that already have functioning enforcement mechanisms, an independent press, organized civil society, and protected political speech. Strip out any of those supporting elements, and the data portals, the open contracting standards, the participatory budgeting apps — they become performance without consequence.

## A Cautionary Note for the AI Moment

The timing of this analysis is pointed. Right now, significant institutional money is flowing toward AI tools designed to strengthen democratic participation and government accountability. Grant programs are offering half-million-dollar awards specifically for democracy-adjacent AI projects. Conferences are convening around local government accountability frameworks. The energy is real.

But if the civic tech sector is about to layer AI capabilities on top of a transparency infrastructure that was already underperforming, the result could be the same disclosure theater at dramatically higher speed and scale. Faster access to data that no one has the capacity to act on is not progress.

What the Ash Center research points toward instead is a model of targeted, context-sensitive intervention — one that asks not just "how do we make this information available?" but "who specifically is positioned to use it, what do they need to act, and what institutional pathways exist for that action to produce change?"

## The Infrastructure Question

For observers of the AI agent economy in particular, there is a structural issue lurking beneath the surface of these findings. Autonomous agents that can query government databases, synthesize budget data, and flag anomalies are genuinely powerful. They lower the cost of investigative work. They can surface patterns that no human analyst scanning spreadsheets would catch.

But they cannot compel a city council to hold a hearing. They cannot protect a whistleblower. They cannot rebuild a local newsroom that closed five years ago. They cannot create the political conditions under which disclosed corruption actually results in consequences for the people who committed it.

Those are human and institutional problems, and they are not amenable to platform solutions.

## What Accountability Actually Requires

The honest takeaway from the Ash Center's work is uncomfortable for a sector that has spent years packaging reform as a technology problem. Accountability requires power — the organized, sustained capacity of citizens, journalists, advocacy groups, and oversight bodies to impose real costs on government actors who abuse public trust.

Transparency can feed that power. AI tools can sharpen it. But neither substitutes for it.

As the next generation of civic AI investments gets made, the question funders, builders, and policymakers should be asking is not how much data they can unlock. It is who has the power to do something with it — and whether that power is being built alongside the tools designed to inform it.