A quiet shift is happening inside government offices, courts, and regulators around the world. People are using AI to write complaints, appeals, and benefit applications that they would have given up on before. The result is a wave of new paperwork that public institutions were not built to handle.
The UK housing ombudsman saw complaints nearly triple in two years, and the US Consumer Financial Protection Bureau logged five times more complaints over the same stretch. Its most recent annual report shows complaint volume more than doubled again in a single year, driven mostly by disputes over credit reports. A new academic paper tracked 84 comparable cases across 11 countries, from Brazil's courts to Germany's parliament, and found the same pattern almost everywhere: flat volumes before 2022, then a steady climb that shows no sign of slowing.
The cause is simple. Filling out a benefit form, writing a legal appeal, or drafting a formal complaint used to require time, confidence, and a bit of know how, and many people with a real claim simply never bothered because the paperwork itself was the obstacle. AI tools remove that obstacle: take a photo of a confusing letter, ask an AI assistant what to do, and it drafts a response in seconds.
This is different from the mess many tech companies faced last year, when bug bounty programs got overwhelmed by AI generated junk reports with no real substance behind them. Several companies had to pause those programs entirely. Public services are seeing something else: mostly real people with real, legitimate claims who are finally able to act on them.
That distinction matters for anyone running a complaints desk, claims department, or customer service line, not just governments. Insurance claims handling, warranty disputes, HR grievance processes, and regulatory complaint lines all work the same way public services do: a form, a human reviewer, and a budget sized for the old volume. If AI makes it this easy to file with a government agency, it will make it just as easy to file a claim, a complaint, or a dispute with any business that has an intake form.
The sensible response is not to treat every new filing with suspicion. Blocking or slowing down submissions punishes the genuine claims along with the rare bad ones, and most of what is showing up is genuine. The better move is to rebuild the intake process itself: clearer forms, faster triage, and systems built to sort real requests from noise at scale, rather than assuming a trickle that a small team can handle by hand.
Volumes like this are not a one time spike. They track the pace at which AI tools keep getting better and cheaper to use, and that pace has not slowed down. Any organization that collects complaints, claims, or applications from the public should plan for a lot more of them, not a return to normal.