Artificial intelligence is dramatically simplifying how people complete forms and submit complaints, leading to massive surges in applications for public services globally. In the United Kingdom, complaints to the housing ombudsman have more than doubled since ChatGPT became available, climbing from 2,600 in 2022 to over 7,000 last year. Meanwhile, the United States' Consumer Financial Protection Bureau (CFPB) experienced a fivefold increase in complaints during the same timeframe. Similar spikes occurred in Brazilian judicial petitions and German parliamentary petitions.

Researcher Chris Schmitz is examining this phenomenon as part of a larger pattern he terms "agentic flooding." In a study scheduled for presentation at the AI Ethics and Society conference next month, he analyzes 84 instances of potential flooding across 11 jurisdictions, uncovering substantial evidence that AI tools are transforming how individuals engage with public services. The more challenging issue is determining the appropriate response. While certain new filings are evidently adversarial, others stem from legitimate applicants leveraging AI to submit claims they might otherwise have abandoned. Although some may view these new applications as AI-generated spam, Schmitz perceives them as a unique chance to overhaul social services for the AI age.

Schmitz's research covered services from welfare applications to formal judicial appeals, all of which offer online access that an AI assistant could utilize. (The complete dataset is available here for reference.) Due to methodological considerations, Schmitz's paper refrains from asserting that AI directly causes the surge in new applicants. Nevertheless, most of the 84 cases exhibit a consistent pattern: submissions remained relatively stable before 2022, then accelerated as AI technology spread. Importantly, most cases show no signs of that growth decelerating, indicating it will likely continue rising for years.

For Schmitz, it's straightforward to understand how improving AI capabilities and accessibility would fuel greater usage. "People are discovering that this is something one can do, and incrementally, it is just getting easier to do it…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," he explained.

This volume increase mirrors what many bug-bounty services encountered last year, when companies found their inboxes overwhelmed with low-quality reports produced by LLMs. These reports seldom identified significant security issues, yet companies were still required to review each submission, creating a considerable resource strain. It's simple to envision public services confronting a comparable challenge, as they handle five times more applicants with unchanged budgets. However, whereas bug bounty programs were inundated with worthless submissions, Schmitz notes that most new applications to public services originate from real individuals with legitimate claims. If these people weren't claiming benefits previously, it might have been because the application process was too daunting—a concept in policy circles known as administrative burden.

Now that AI can alleviate that burden, it presents an opportunity to redesign many of these services in a more AI-compatible manner. This is a significant undertaking that has largely not been addressed, but Schmitz views it as grounds for optimism. "A big part of making AI go well is being able to detail out what the good version of things looks like. And anyone who's ever used ChatGPT to do the tax return knows that there's a good version here where you're being helped," Schmitz says. "This could be the moment to say, 'we need to rethink pretty much everything about how this process looks.'"