The public is more engaged than ever. People want to know how their tax dollars are spent and how policy gets made, and they have every right to ask. But a new reality is making it harder for local governments to keep up: a flood of public records requests generated by AI.
Generative AI has made it effortless to fire off massive, sprawling records requests with a single click. And this is no longer just individuals. Entire companies have now built teams whose only job is to submit public records requests at scale, then resell that information to vendors and the private sector. For the government staff on the receiving end, the result is a crushing workload.
Faced with that workload, lawmakers are moving to find solutions. California's Assembly Bill 1821, authored by Assembly member Blanca Pacheco, is one early example. The bill would require proper submission methods to be defined by each agency, establish fees for commercial-use requests and institute penalties for requests submitted with malicious intent. Most importantly, it would extend response timelines.
California is not alone. As open records requests surge nationwide, agencies everywhere are weighing how to recover the cost of mass fulfillment. Every hour spent on records requests is an hour pulled from other essential work: processing business permits, maintaining public archives, running elections.
The numbers are staggering.
A single request in Chula Vista, California required 150 to 300 staff hours. In Fontana, one person filed more than 100 requests with the stated goal of disrupting operations, costing the city over $300,000 in legal and staffing expenses.
Versions of that story are playing out in communities across the country. Call it what it is: administrative warfare, waged one records request at a time.
Cost and burnout are only part of the story. The bigger risk is compliance.
Public records laws don't just encourage agencies to respond. They require it, on a clock. Under the California Public Records Act, an agency generally has 10 days to determine whether it will produce records, and similar deadlines exist in nearly every state. Those timelines were written for a world of paper files and the occasional request, not for an inbox flooded by AI.
When volume outruns staff, deadlines start slipping. And a missed deadline is not a minor administrative hiccup. It is legal exposure. Requesters who sue to force production and prevail are routinely awarded attorney's fees, so a single backlog can become a court order plus a legal bill the taxpayers ultimately pay.
Bills like AB 1821 are an understandable response. More time is better, right? But there is a faster, more effective fix.
The same technology driving up records volume can also resolve those requests at a fraction of the time usually spent. We can close the operational gap with better tools.
That is exactly what we built at Madison AI. Our public records fulfillment model safely retrieves, sorts, and redacts municipal records, turning a workflow that once took several days into one that takes about an hour.
It handles the heavy lifting end to end: automated intake tagging, near-duplicate detection, AI-assisted redaction, and complete audit trails that keep agencies compliant with retention and privacy laws.
The preliminary results are striking. In Washoe County, Nevada, our development partner, the public records assistant is delivering up to 90% time savings, cutting as much as 9 to 10 hours of staff work off a single request. Palm Springs and Culver City are rolling out the same model next.
And this is not a one-size-fits-all product. We tailor each assistant to an agency's specific workflows and data, built alongside the people who live these operational realities every day. The same approach works for a small town or a large county, in any state.
At the end of the day, everyone wants the same thing: a government that is open, accountable, and transparent. So we urge lawmakers and city leaders, in California and beyond, to consider different tools rather than extended timelines.
What if a state created a safe harbor for agencies that adopt an AI-first response model? What if requesters used an automated portal before submitting follow-ups? What if agencies that deployed credible AI platforms, met basic public records criteria in good faith, and resolved the bulk of public inquiries earned more latitude on response timing and cost recovery?
That approach rewards transparency. Let's uphold the spirit of open government and give local governments the technology to make transparency easy.
