AI-Assisted Filings Flood Government Agencies With Legitimate Claims

Public agencies worldwide are experiencing an unprecedented surge in applications, complaints, and formal petitions, driven directly by the expanding capabilities of modern artificial intelligence tools. As AI systems grow more sophisticated and accessible, they are systematically lowering the friction that once prevented ordinary citizens from engaging with government institutions. The result is overwhelming: public services designed for predictable volume are struggling to absorb submissions arriving at three, five, and even ten times historical rates.

The documented surge across continents and systems

United Kingdom housing complaints more than doubled

The United Kingdom housing ombudsman, an independent body that reviews complaints about housing associations and landlords, provides the clearest numerical picture. The agency received just over 7,000 complaints last year, up from 2,600 in 2022—a 170 percent increase that tracks almost precisely with ChatGPT’s launch and mainstream adoption. The timing is not coincidental. Submitting a formal complaint to a housing ombudsman requires specific language, documentation of previous attempts to resolve the issue, and understanding of procedural requirements. For many residents dealing with housing problems, that barrier proved insurmountable. Now, AI tools can draft the complaint automatically.

American financial regulator sees five-fold growth

The United States Consumer Financial Protection Bureau, a federal agency that tracks consumer complaints about banks and financial institutions, recorded a five-fold increase over the same 2022 to 2025 period. That agency handles complaints about lending, credit reporting, debt collection, and other financial practices that directly affect household finances. A five-fold increase means the CFPB’s complaint database, which was already substantial, has become a genuine flood. Brazil’s judicial system documented comparable increases in formal court petitions. Germany’s parliamentary petition system—the mechanism by which citizens formally petition their government—recorded parallel growth. The pattern repeats across jurisdictions serving different populations, different legal systems, and different types of public services, all pointing to a single underlying cause: AI is changing how people engage with bureaucracy.

Schmitz’s research on “agentic flooding”

Read also: Maven Robotics Emerges With $100M to Deploy Warehouse Robots · Paul Christiano joins OpenAI board as safety scrutiny intensifies · Why AI-Generated Food Photography Triggers Visceral Disgust

Researcher Chris Schmitz has undertaken the most comprehensive analysis of this phenomenon. His work, documented in a paper scheduled for presentation next month at the AI Ethics and Society conference, examined 84 documented instances of potential flooding across 11 jurisdictions. The research spans public services ranging from welfare benefits applications to formal judicial appeals—essentially any government-facing system that operates online and can be accessed or assisted by an AI tool. Schmitz’s team compiled data on submission volumes, approval rates, timeline, and technological context for each case, building what may be the first rigorous account of how AI is changing citizen-government interaction at scale.

The growth pattern shows no signs of stopping

The data reveals a striking historical break. Across nearly all the 84 services Schmitz studied, submission volumes remained stable or grew slowly before 2022. Since then, growth has accelerated significantly. For methodological reasons, Schmitz’s published paper stops short of making causal claims—it is difficult to isolate AI as the sole cause of any single increase. However, the pattern across 84 independent cases is difficult to explain without reference to AI. Most significantly, growth has not slowed. If the surge were a temporary novelty effect, volumes would be expected to plateau or decline. Instead, Schmitz’s data shows sustained acceleration, suggesting that these levels will persist and likely increase for years to come.

AI made the process dramatically simpler and faster

The mechanism driving adoption is straightforward. Before AI tools became commonplace, applying for benefits, filing a complaint, or submitting a formal petition required applicants to gather context, research procedural requirements, draft appropriate language, and navigate online submission systems. Schmitz describes the before and after this way: “Before it might have been a question of a lot of dragging context together and prompting ChatGPT 3.5 very precisely, it may now be a question of just pasting or taking a photo of a letter with your Claude app and getting a pretty good response in one shot.” That shift from multi-step, high-friction process to single-step, nearly automatic response matters enormously. As more people become aware the tool exists and how simple it is to use, adoption accelerates.

Separating legitimate claims from generated spam

Bug-bounty platforms flooded with AI-generated noise

This surge is not unprecedented, but the context matters significantly. Last year, bug-bounty platforms—services that pay researchers for reporting security vulnerabilities—experienced comparable flooding. Security researchers using AI to generate vulnerability reports submitted hundreds of mostly worthless reports to companies. Companies were legally and contractually obligated to review each submission. The result was a genuine plague of noise: low-quality, algorithmically-generated junk consuming resources that could have been spent on legitimate security work. Agencies facing similar flooding in the public sector initially feared identical outcomes.

However, Schmitz’s analysis reveals a crucial difference. While bug-bounty platforms received floods of worthless submissions, most new applications to public services come from real people making genuine claims. This is the finding that reframes the entire problem. Schmitz put it directly: “The vast majority of cases we find are people who are entitled to claim for something, claiming for that thing.” These are not adversarial actors or bots trying to game the system. They are citizens who qualify for benefits, who have legitimate housing complaints, or who have standing to file legal petitions—but who previously abandoned the effort because the procedural burden was too high. AI did not make them eligible. It made them willing to try.

Operational strain on government agencies

The resource math is unforgiving. An agency that saw five times more complaints with the same staff budget faces a five-fold increase in work per employee. Staffers who previously had time to thoroughly investigate each complaint now triage by priority, leaving lower-level cases to languish. Processing times extend from weeks to months or years. For agencies serving vulnerable populations—housing ombudsmen work with renters often in precarious situations, financial regulators handle complaints from people harmed by predatory lending—service delays are not merely inconvenient; they can have serious real-world consequences. Citizens waiting for housing complaints to be addressed or for financial regulator action to freeze fraudulent transactions are suffering ongoing harm while the backlog clears.

Public agencies rarely have the flexibility to rapidly expand their budgets. The surge of legitimate claims arrives faster than institutional capacity can grow. The result is unsustainable pressure: either service quality drops as cases are rushed through, or wait times become prohibitive.

The opportunity to redesign public services fundamentally

Rather than accepting the surge as a crisis to be endured, Schmitz proposes reframing it as an opportunity to redesign public services for the AI era. The current systems were architected decades ago when manual review, paper forms, and human data entry were unavoidable constraints. Those constraints no longer apply. When an individual successfully uses AI to prepare a tax return or draft a formal government application, the result is often better than what they would have produced alone: clearer language, more complete information, better-organized evidence. The process is genuinely improved for both the applicant and the reviewing agency.

Reimagining public services around AI assistance would mean building systems that expect and accept AI-generated submissions, that format requests to make AI processing easier, that automate routine tasks currently handled by human staff, and that design interfaces assuming users may be getting help from AI tools. This would be a major undertaking, one that most agencies have barely begun to contemplate. Yet for Schmitz, the challenge represents the core work of making AI governance succeed. “A big part of making AI go well is being able to detail out what the good version of things looks like,” he said. “Anyone who’s ever used ChatGPT to do the tax return knows that there’s a good version here where you’re being helped. This could be the moment to say, ‘we need to rethink pretty much everything about how this process looks.’”

That moment is now. The surge of legitimate claims will not reverse. AI capability will continue improving. More citizens will discover and use these tools. Public agencies can resist that trend and watch service quality deteriorate, or they can recognize this as the window for structural change—the chance to build systems that handle both human and AI-assisted applicants seamlessly. The cost of waiting is perpetual crisis. The potential benefit of acting is fundamentally better service for citizens.