AI agents are quietly changing how millions of people interact with government — and the numbers are already turning heads around the world. 🌍
In the United Kingdom, complaints to the housing ombudsman have more than doubled since ChatGPT arrived, jumping from 2,600 in 2022 to just over 7,000 last year. In the United States, the Consumer Financial Protection Bureau (CFPB) saw its complaints grow fivefold over the same period. Brazil and Germany have also seen similar spikes in court petitions and parliamentary petitions, respectively.
What is behind all of this? The answer is straightforward: it has become much easier to fill out forms, file complaints, and access public services with the help of artificial intelligence. And based on the data, this shift is just getting started.
Researcher Chris Schmitz has been tracking this phenomenon closely, and what he has found might make you rethink the role AI plays in everyday life. A significant portion of these new requests are not coming from bots trying to game the system — they are coming from real people with legitimate claims who previously just gave up when faced with red tape.
But are public services ready to handle this volume? That is exactly what we are going to dig into. 👇
What the data is telling us
When numbers start climbing this dramatically across such different countries at the same time, it is hard to chalk it up to coincidence. Schmitz coined the term agentic flooding to describe this wave of activity driven by AI agents. In a paper set to be presented next month at the AI Ethics and Society conference, he analyzed 84 distinct cases of potential flooding across 11 jurisdictions and found widespread evidence that AI tools are reshaping how people engage with public services.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
What the data shows, in practical terms, is a structural shift in how people deal with administrative burden — that heavy, exhausting feeling of having to wade through endless forms, confusing legal language, and processes that seem designed to discourage anyone who actually needs help. For decades, many people simply gave up halfway through, not because the right did not exist, but because the effort required to exercise it was just too much.
It is worth noting one methodological detail here: Schmitz’s study does not claim that AI is directly causing this increase in requests. Even so, the majority of the 84 cases follow the same pattern. Submissions were essentially flat before 2022 and then began climbing at an increasingly rapid pace as AI technology spread. And here is the most interesting part: in most cases, this growth has shown no signs of slowing down, which suggests it could keep rising for several more years.
With the arrival of AI agents, this landscape has started shifting for good. Tools that once required technical knowledge or access to lawyers and specialized consultants are now available to anyone with a smartphone and an internet connection. An AI agent can read a regulation, identify the right form, fill in the required fields, and even draft a formal complaint letter — all within minutes, using language appropriate for the context.
Schmitz himself explains how this ease of use fuels adoption. According to him, people are discovering that this is something they can actually do, and every day it gets simpler. Before, you had to pull together a lot of context and write very precise prompts for ChatGPT 3.5. Today, you can often just paste some text or snap a photo of a letter through the Claude app and get a thorough response right away. The direct result is exactly what we are seeing in the data: a surge in requests that, in some cases, has tripled or even quintupled the volume public agencies were used to processing.
Legitimate claims vs. real-world overload
Here is an interesting parallel worth drawing. The jump in volume looks a lot like what many bug bounty programs — the ones that reward people for finding security flaws — experienced last year. Their inboxes were flooded with low-quality reports generated by language models. Most of the time, these reports did not contain any meaningful security vulnerabilities, but the companies still had to review each one, which turned into a massive drain on resources. It is easy to picture public services facing a similar problem, having to handle five times the number of requests on the same budget as always.
The big difference, though, lies in the nature of these requests. While bug bounty programs were swamped with worthless submissions, Schmitz observes that most of the new requests to public services come from real people with legitimate claims. In his words, the overwhelming majority of cases involve people who have a right to claim something and are, finally, claiming it.
Think about someone living in a property with serious structural problems who never knew how to file a formal complaint with the housing ombudsman. With an AI agent walking them through the process, that person can finally do what they should have done years ago — and the system, at last, has to respond. If these people were not claiming their benefits before, it was probably because the work involved in applying felt too intimidating.
This raises an important question: if the system was functioning under artificially low demand because of bureaucracy, the current volume might be closer to the actual demand than any previous estimate. Public agencies, in many cases, sized their teams, processes, and infrastructure based on a scenario that underestimated how much support people truly needed. Now that the barrier to entry has dropped, the gap between installed capacity and real need is becoming visible — and it is big.
Still, it is only fair to acknowledge that not every request generated with the help of AI comes with the same quality of information or relevance. There are cases where the ease of access leads to incomplete filings, misdirected claims, or submissions that require far more human effort to process than ideal. The challenge for public services, then, is not just handling more volume — it is handling a larger volume while making sure every case gets the attention it deserves, without letting efficiency collapse alongside response capacity.
A rare opportunity to rethink services
This is precisely where Schmitz sees reason for optimism. Rather than treating these new requests as just AI-generated spam, he prefers to see this moment as a rare opportunity to reinvent social services for the age of artificial intelligence. After all, the administrative burden that has always weighed on citizens now has a real chance of being reduced.
For him, a fundamental part of making AI work well is being able to articulate what the good version of things looks like. And anyone who has ever used ChatGPT to help with their tax return knows there is a good version of this story, one where you are genuinely being helped. Schmitz believes this could be the moment to say we need to rethink practically everything about how these processes work.
It is worth highlighting that the cases he analyzed are quite varied. They range from social benefit applications to formal legal appeals, but they all share one thing in common: they involve online services that can be accessed by an AI assistant. It is this characteristic that opens the door to automation — both on the side of the person making the request and on the side of whoever needs to respond to it.
What changes for public services from here
The question many public administrators are starting to ask is: how does the government adapt to a reality where the administrative burden on the citizen side has dropped dramatically, but on the government side it remains the same? For a long time, the complexity of processes served, in a way, as a natural filter — not intentional, but very real. Now that technology has removed that filter, the pressure falls squarely on those who need to process, analyze, and respond to every request that comes in.
Some countries are already discussing the adoption of AI agents on the public side as well — tools that help with initial screening of requests, identify priority cases, automate responses for simpler matters, and free up human workers to focus on what truly requires judgment and empathy. This approach makes sense, but it carries its own risks: automation without proper oversight can lead to flawed decisions that directly affect people’s lives, especially the most vulnerable, who are precisely the ones who have benefited the most from the surge in requests enabled by AI.
What becomes clear, looking at the bigger picture, is that we are witnessing a transformation with no turning back. AI agents are already part of daily life for a growing segment of the population, and that segment is only going to get bigger. Public services that manage to adapt to this new pace — whether by investing in human capacity, adopting technology responsibly, or redesigning processes to be more efficient — will come out ahead. Those that ignore this shift will find themselves increasingly buried under a volume of demands that the current model simply cannot absorb. And caught in the middle of all this are the citizens waiting for an answer.
At the end of the day, the message is clear: artificial intelligence is not just making life easier for people who need help. It is exposing, in a very transparent way, the true size of the demand that has always existed behind the wall of bureaucracy. And it is now up to governments to decide whether they will treat this as a problem to be contained or as the chance to build public services that are more human, accessible, and efficient. 💡
