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

NewsOnScale Staff

July 14, 2026

For two decades, the civic tech movement built its cathedral on a simple premise: if citizens can see what government is doing, they can hold it accountable. Open data portals, public meeting archives, budget dashboards, FOIA pipelines — the infrastructure of sunlight was supposed to be enough. A growing body of evidence says that premise was always incomplete, and in some cases, dangerously so.

Research emerging from the Ash Center at Harvard Kennedy School is among the clearest recent articulations of this problem. The core finding is not surprising to anyone who has spent time in government accountability work, but it remains underappreciated in technology circles: transparency produces accountability only when certain conditions are already in place. Without functioning civil society organizations, a capable press, institutional capacity to act on disclosed information, and political will to enforce consequences, raw data disclosure tends to benefit the people who already have power rather than those who don't.

## The Information Gap Is Not the Only Gap

The civic tech community got seduced by a particular theory of change. If corrupt officials hide their activities, exposure stops corruption. If agencies bury their budgets, publication creates accountability. The logic is clean. The problem is that it treats information asymmetry as the only asymmetry that matters.

In practice, power asymmetries matter just as much. A community without legal resources, time, or organizational infrastructure cannot convert a disclosed procurement record into a lawsuit, a news story, or a policy change. A well-resourced interest group absolutely can. This means transparency tools, absent complementary investments, can function as a subsidy to sophisticated actors — lobbyists, contractors, political operatives — who know exactly what to do with raw government data and have the resources to do it.

This is not an argument against transparency. It is an argument against transparency as a complete strategy.

## The AI Amplification Problem

The stakes of getting this right are rising fast. Artificial intelligence tools are beginning to integrate into civic data infrastructure — summarizing council meetings, flagging budget anomalies, parsing regulatory filings at scale. Philanthropic capital is flowing into this space. ICTworks is currently circulating a $500,000 grant opportunity specifically aimed at AI-assisted democracy projects.

If the underlying theory of change is flawed, AI doesn't fix it — it scales it. Faster, cheaper, more comprehensive disclosure of government data still requires the same downstream conditions to translate into accountability: organized citizens, responsive institutions, enforceable consequences. An AI that can surface a contracting irregularity in seconds is only useful if someone with standing and capacity picks it up.

The risk is that AI-powered transparency tools become an even more impressive-looking version of the same incomplete solution — generating metrics of openness that substitute for genuine accountability infrastructure.

## What Accountability Actually Requires

The harder work involves building the intermediary layer that converts disclosed information into action. That means investing in local journalism, not just data portals. It means funding community organizations with the capacity to interpret and act on government disclosures. It means designing enforcement mechanisms that don't require citizens to bear the entire burden of follow-through.

It also means being honest about what governments are actually disclosing. Several municipal governments around the world have published open budget data that is technically accessible and practically uninterpretable — aggregated to the point of meaninglessness, formatted in ways that require specialist knowledge to parse. This is transparency as performance rather than transparency as function.

## The Accountability Stack

NewsOnScale has consistently covered the AI agent economy with attention to who actually benefits when new tools enter legacy systems. The civic technology lesson applies directly: the question is never just whether a tool creates more openness. The question is who has the capacity to use that openness, and whether the institutions responsible for responding to it are designed to do so.

Transparency is necessary. It is not sufficient. The civic tech movement is learning this slowly, and the AI layer arriving on top of broken accountability infrastructure will learn it too — unless the field asks harder questions now, before the capital is already deployed.