AI agents are making it easier for the public to interact with government services, helping users apply for benefits, understand policies, and voice opinions. However, these surges in demand—termed 'agentic flooding'—could strain unprepared government systems. Based on a dataset of 84 potential cases across 11 jurisdictions, researchers find that large language models (LLMs) are likely driving widespread flooding through cheap text generation.

High-risk services include financially attractive but complex ones. The study proposes a risk matrix to assess exposure and maps potential government responses. While historical precedent suggests responses will likely contain most flooding, rapid measures like fees may limit equitable access.

The authors recommend near-term actions to mitigate flooding without compromising access.