Civic
A new analysis from Harvard's Ash Center argues that making government data visible was never the same thing as making government answerable.
NewsOnScale Staff
August 9, 2026
For more than a decade, the civic technology movement ran on a simple and seductive premise. If citizens could see what their governments were doing — budgets, contracts, voting records, permit approvals — accountability would be a natural consequence. Sunlight, the old saying went, is the best disinfectant. Build the dashboard, and the disinfectant would flow.
A new analysis from Harvard Kennedy School's Ash Center for Democratic Governance and Innovation challenges that premise with uncomfortable directness. Transparency, the researchers argue, is not accountability. It is, at best, a precondition for it — and treating the two as interchangeable has quietly undermined a generation of reform efforts.
## The Gap Between Visible and Answerable
The distinction matters more now than it ever has. As AI tools begin entering the civic space — promise-laden platforms for parsing government records, flagging anomalies in public spending, and surfacing patterns in enforcement data — the same foundational error risks being replicated at scale. If the first generation of civic tech built portals that no one used to hold anyone accountable, the second generation risks building smarter portals with the same structural blind spot.
What the Ash Center analysis surfaces is something practitioners in this space have quietly known for years: data publication without enforcement pathways, without institutional capacity to act on findings, and without communities equipped and motivated to engage, produces very little systemic change. A spending database that reveals a suspicious pattern of contract awards is only useful if a journalist investigates it, a regulator can act on the complaint, and a political system has sufficient independence to punish the wrongdoing. Remove any one of those links, and the transparency moment evaporates.
This is not an argument against open data. It is an argument for being precise about what open data can and cannot do on its own.
## Accountability Requires Infrastructure, Not Just Information
The lesson applies with particular force to the AI accountability tools now entering the market. FactCheck Africa recently unveiled an AI-driven platform designed to monitor government claims and flag misinformation in public communications — a meaningful effort in a context where official opacity is a well-documented governance failure. But the platform's long-term impact will depend on factors the technology itself cannot supply: legal protections for journalists who use its findings, political will to sanction officials caught in contradictions, and public trust in the institutions meant to respond.
Similarly, a $500,000 grant program now accepting applications for AI projects aimed at strengthening democratic participation represents genuine investment in a field that needs it. But grant-funded pilots rarely survive long enough to stress-test the accountability infrastructure around them. The technology gets built; the surrounding ecosystem of enforcement, civic capacity, and political incentive often does not.
## What a Corrected Framework Looks Like
The Ash Center's corrective is less a demolition of civic tech than a maturation of it. The field's next phase needs to ask harder questions before shipping. Who is the actual accountability actor in this system — the voter, the journalist, the regulator, the court? What mechanism converts a data finding into a consequence? And who bears the cost when the platform identifies a problem that no institution is willing to address?
These are not technology questions. They are political and institutional ones, and the AI agent economy now building tools for civic use would do well to treat them as design requirements rather than afterthoughts.
Transparency tools that lack a theory of accountability are, in the end, documentation projects. They create a record of what went wrong. That is valuable. It is not sufficient.
The disinfectant metaphor was always a little too clean. Sunlight reveals; it does not automatically compel. Building the infrastructure that converts revelation into remedy is the harder, slower, less fundable work — and it is the work that actually determines whether any of this matters.