What Fowler found and how large it was
Security researcher Jeremiah Fowler discovered an open, unpassworded database tied to CitySourced, a US civic technology company whose software allows residents to report local problems such as potholes, broken streetlights, and code violations directly to city authorities. The exposed database contained roughly 42 million documents alongside 450,000 JSON files, a scale that places this firmly among the larger unsecured-database discoveries reported this year rather than a minor, isolated misconfiguration.
Most of the exposed records consisted of images attached to resident complaints or code violation reports, the kind of visual documentation residents submit when flagging an issue to their local government. Embedded within that image data was personally identifiable information in plaintext, including names, postal and email addresses, and phone numbers, information that residents submitted expecting it to stay within a municipal reporting system rather than sit openly accessible online to anyone who happened to find the unsecured server.
The detail that moves this past a routine misconfiguration
Beyond the resident records, Fowler found thousands of internal records spanning a full decade, from 2016 to 2026, suggesting the exposed environment had accumulated data over a long operational history rather than reflecting a recently created, narrowly scoped database. Directory names such as Api, Customers, Data, and Logs pointed to a broader internal system rather than an isolated test environment that might otherwise explain away the exposure as low-stakes and easily contained.
More concerning, Fowler identified .sdk and .exe files within the exposed directories, which he said could indicate an attempt to plant malware inside the environment. That detail shifts the story from a straightforward data privacy failure toward a potential active compromise, since an unsecured database containing executable files is a plausible staging ground for an attacker rather than simply a passive information leak sitting unnoticed, and it changes what kind of response the discovery actually calls for.
Who actually owns the problem is not entirely clear
Fowler was unable to determine definitively whether CitySourced itself operated the exposed database or whether a third-party contractor managed it on the company's behalf. That ambiguity is a familiar pattern in civic technology, where a single resident-facing application frequently depends on a chain of vendors and subcontractors handling hosting, data processing, and backend infrastructure, any one of which can introduce the specific configuration failure that exposes the data to the open internet.
CitySourced's own corporate history adds another layer to that ambiguity. The company merged with Rock Solid Technologies in 2019 and was acquired by Granicus roughly three years later, a sequence of ownership changes that can easily leave legacy systems and their original security configurations unclear to the current owner, particularly for infrastructure that predates the most recent acquisition and was never fully audited afterward by the team now responsible for it.
How the company responded once notified
After Fowler contacted CitySourced, the database was secured. A company statement thanked Fowler for the responsible disclosure and for following up, and Granicus said it had reviewed the information he provided and investigated the matter internally. That response follows the standard playbook for handling a responsible disclosure, acknowledgment, remediation, and an internal review, though the public statements provided do not clarify how long the database sat exposed before Fowler's discovery.
That gap matters because the duration of an exposure window is often the single most important variable in assessing actual harm, far more than the raw record count alone. A database exposed for days carries meaningfully less risk than the same database exposed for months or years, and without that detail, residents whose information appeared in this database have no way to gauge how much real exposure risk they actually faced.
Why civic technology data deserves the same scrutiny as corporate data
Resident complaint data might seem lower-stakes than financial or health records at first glance, but the personal information attached to it, home addresses tied to specific complaints, phone numbers, and years of accumulated internal records, carries real risk when exposed at this scale. A resident who reported a neighbor's code violation or a personal safety concern through a civic app reasonably expects that submission to stay within a controlled municipal system rather than sit in a publicly accessible database.
This incident is a useful reminder that civic technology vendors, often smaller companies operating with less security scrutiny than major enterprise software providers, sit at the center of an enormous amount of personally identifiable data flowing between residents and local government. Cities and counties procuring this kind of software should be applying the same vendor security due diligence they would apply to any enterprise data processor, not a lighter standard simply because the vendor serves the public sector.
What this means for your own vendor data governance
If your organization works with civic technology vendors, government data processors, or any multi-layered vendor stack where responsibility for a specific database's security configuration is not clearly assigned in your contracts, this incident is a prompt to close that gap now rather than after a similar exposure. Explicitly document which party owns security configuration for every system touching resident or customer data, including subcontractors several layers removed from your direct vendor relationship.
More broadly, treat the presence of executable files inside any exposed data environment as a signal to investigate for active compromise, not just a data privacy incident to remediate and close. Fowler's finding here is a reminder that a misconfigured database is not always a passive exposure, it can also be infrastructure an attacker has already started using, and your incident response process should account for that possibility from the first moment such a database is discovered.



