Agentic Flooding: Researchers Name the Surge in AI Assisted Government Filings

A paper due at the AAAI Conference on AI, Ethics and Society in October gives a name to something governments are already dealing with: agentic flooding, the surge in demand that follows when AI makes it cheap to apply, complain or petition. Chris Schmitz, Lewis Hammond and Alan Chan collected 84 potential cases across 11 jurisdictions and conclude that flooding is likely happening widely today, mostly through large language models generating text cheaply.

The authors are careful about what the dataset shows. They posit that flooding is occurring rather than claiming it as established, and the paper's own limitations section covers both the collection methodology and the identification of risk factors.

The exposure analysis is the useful part for anyone running a service. The paper develops a risk matrix and finds near term risk highest for services that are financially attractive but complex, which is the combination that makes an AI assistant worth pointing at a form.

On what governments can do, the paper maps responses into suppressing demand and increasing capacity, and notes that precedent suggests these will be enough to stop most cases. The warning is about which response arrives first. Friction inducing measures such as fees are the fastest to deploy and they trade off equitable access to public services, which falls on the applicants who were least able to file without help. The authors close on three near term recommendations intended to avoid that trade: auditing exposure, developing a digital identity strategy, and establishing legal certainty.