LLM-generated benefit appeals are flooding government services

Public services are increasingly strained by LLM-written appeals for benefits

A new study warns that AI agents using large language models are flooding government services with cheaply generated appeals, straining unprepared agencies. Analyzing 84 potential cases across 11 jurisdictions, the authors find flooding is already widespread, with financially attractive but complex services most at risk. They propose a risk matrix and map possible responses, cautioning that quick fixes like fees may undermine equitable access. The paper recommends near-term actions to mitigate flooding without sacrificing fairness.