Complaints to the UK housing ombudsman rose from 2,600 in 2022 to more than 7,000 last year, and the US Consumer Financial Protection Bureau saw a 5x increase in complaints over the same window. The trigger, according to a new paper from researcher Chris Schmitz, is ChatGPT and the class of consumer AI assistants that followed it. Schmitz calls the pattern 'agentic flooding' — a measurable surge in citizen filings driven by AI tools that draft the paperwork.
Schmitz's paper, set for presentation next month at the AI Ethics and Society conference, examines 84 different cases of potential flooding across 11 jurisdictions. The dataset spans welfare applications, judicial appeals, parliamentary petitions, and consumer-finance complaints. Brazilian judicial petitions and German parliamentary petitions show the same shape: roughly flat volumes before 2022, then a rising curve that has not yet flattened.
The paper stops short of a direct causal claim on methodological grounds, but the arc is consistent across jurisdictions. And the mechanism is intuitive — each generation of consumer AI has cut the effort required to draft a formal complaint from hours to minutes.
Key facts
- 01UK housing ombudsman complaints rose from 2,600 in 2022 to over 7,000 last year, more than doubling since ChatGPT's release.
- 02The US Consumer Financial Protection Bureau saw 5x growth in complaints over the same period.
- 03Researcher Chris Schmitz documented 84 cases of application surges across 11 jurisdictions in an upcoming AI Ethics and Society paper.
- 04Similar jumps appeared in Brazilian judicial petitions and German parliamentary petitions, with volumes still climbing.
- 05Most filings come from legitimate claimants, not spam — AI is lowering the administrative burden that kept people from applying.
The rise mirrors what bug-bounty programs experienced last year, when companies found their inboxes flooded with low-quality LLM-generated vulnerability reports that still had to be triaged. Public agencies now face a version of that same volume problem, but with a critical difference in the underlying signal quality.
Where the bug-bounty flood was mostly noise, Schmitz argues the public-services flood is mostly signal. His paper concludes that the new filings are largely from real people with real entitlements — claims that would previously have been abandoned because the paperwork was too forbidding.
“The vast majority of cases we find are people who are entitled to claim for something, claiming for that thing.”— Chris Schmitz, AI researcher
The policy term for that friction is administrative burden: the time, complexity, and confusion that keep eligible people from claiming benefits they qualify for. AI assistants like Claude and ChatGPT now let a user photograph a letter and get a serviceable draft response in a single prompt. That collapse in effort is what turned a trickle of applicants into a flood.
For overstretched agencies, the near-term picture is grim: five times more applicants against the same budget, with the same statutory obligation to review each filing. But Schmitz frames the moment as an opening to redesign services around AI-assisted intake rather than treat every new filing as spam to be filtered out.
There are limits to the optimism. Some filings are adversarial, and agencies still need triage systems that can distinguish a legitimate benefit claim from a template-generated nuisance complaint. The paper does not resolve how agencies should staff or fund the response, and most jurisdictions have not begun the redesign work Schmitz is calling for.
The volumes also keep climbing. Most of the 84 cases show no sign of the growth curve flattening, which suggests agencies designed for pre-2022 application rates will remain under structural pressure for years. Whether that pressure produces reform or backlog depends on choices governments have not yet made.
The story worth watching is which agencies treat agentic flooding as a workload crisis versus a delivery opportunity. Governments that build AI-native intake — structured forms an agent can fill correctly, machine-readable eligibility rules, automated first-pass review — capture the upside of citizens finally claiming what they are owed. Governments that respond with tighter filters and longer queues will spend the next decade watching legitimate claimants get squeezed out by a bottleneck of their own making. The technology is not going to slow down to give them time to decide.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
