AI-assisted filings are increasing pressure on public services
AI tools are reshaping demand for public services
Public agencies in several countries are receiving substantially more complaints, petitions and applications as generative AI makes official forms and claims easier to prepare. In the United Kingdom, complaints to the housing ombudsman increased from 2,600 in 2022 to just over 7,000 last year after ChatGPT’s introduction. Over the same period, the US Consumer Financial Protection Bureau saw complaint volumes grow fivefold.
Similar increases have appeared in Brazilian judicial petitions and German parliamentary petitions. Researcher Chris Schmitz describes the wider pattern as “agentic flooding”: AI-enabled interactions that rapidly increase the volume of requests arriving at institutions with online processes.
Evidence across 11 jurisdictions
Schmitz examined 84 potential cases across 11 jurisdictions for a paper scheduled for presentation at the AI Ethics and Society conference next month. The cases span welfare applications, official judicial appeals and other public-facing services that can be accessed online by an AI assistant.
For methodological reasons, the paper does not conclude that AI directly caused each increase. However, Schmitz found a recurring trajectory: submission levels were broadly flat before 2022, then rose at an accelerating rate as AI tools became more widely available. In most cases, that growth has not yet slowed.
Schmitz links the trend to the declining effort required to assemble context and draft a response. A task that once demanded careful prompting of ChatGPT 3.5 can now involve pasting text or photographing a letter in an app such as Claude and receiving a usable draft in one step.
Capacity strain and a possible access gain
The pattern resembles the rise in low-quality, LLM-generated reports submitted to bug-bounty programmes. Those programmes still had to assess each report, even when few contained significant security issues, creating a substantial operational burden.
Public services may face a similar capacity challenge if they must process several times as many requests without additional budget. Yet Schmitz argues that many public-service filings are not spam: they are submitted by people entitled to make a claim who previously gave up because the application process was too difficult.
That difficulty is often described in policy as administrative burden. If AI reduces it, higher submission volumes may indicate improved access as well as greater strain. The practical implication for organisations is to redesign intake, validation and case-management processes so legitimate AI-assisted requests can be handled at higher volume without treating every automated draft as inherently suspect.