Schmitz Documents 84 Cases of Agentic Flooding in 11 Nations

Complaints to the UK housing ombudsman more than doubled since ChatGPT arrived, rising from 2,600 in 2022 to just over 7,000 last year. The US Consumer Financial Protection Bureau saw fivefold growth over the same period. Brazilian judicial petitions and German parliamentary petitions followed the same curve.

Chris Schmitz, a researcher at the Centre for the Governance of AI and the Hertie School, calls the pattern "agentic flooding." His team, in a paper posted to arXiv in August and previewed by TechCrunch, screened 2,288 candidate services across twelve countries and pulled 84 cases across eleven jurisdictions where request volumes surged for reasons tied to AI. In 58 of them, government officials themselves attributed the increase to AI.

The point Schmitz keeps pressing is that this is not a spam attack. "The vast majority of cases we find are people who are entitled to claim for something, claiming for that thing," he told TechCrunch. Justice and legal services account for 23 percent of the flagged cases, regulatory complaints for 12 percent, benefits and social protection for 11 percent.

The arc is consistent. Submissions were roughly flat before 2022, then began climbing at increasing speed as language models diffused, and in most cases the growth has yet to slow. Fourteen of the 84 cases already show governments responding with added friction. Thirteen show full service redesigns. Twenty-one show agents deployed inside the processing pipeline itself.

Schmitz will present the dataset at the AI Ethics and Society conference next month. It lands in a run of stories about agents showing up in places they weren't invited, though this one flips the register: the agents are working for the citizens.