New EPIC Report Delves into State AI Procurement

Today, EPIC published a new report detailing the risks of AI systems purchased and used by state governments. Across the country, state and local agencies are outsourcing important government decisions to private companies and their AI tools—all without meaningful public input or oversight. These systems assign children to schools, inform medical decisions about patients, impact policing decisions about where to patrol and whom to target, determine who receives public benefits, and more. And they are all developed and operated by private companies like Deloitte, Thomson Reuters, and LexisNexis.

EPIC’s report, Outsourced & Automated: How AI Companies Have Taken Over Government Decision-Making, builds on two years of research into state AI procurement and includes insights gleaned from 621 different contracts. Part One of the report describes the various risks that private AI systems pose when used for government decision-making, including data privacy risks, issues involving AI bias and reliability, and undermining government accountability. Part Two traces the AI vendor landscape across the country, highlighting three procurement processes that vendors can use to embed their AI systems within state agencies and ten vendors that take the lion’s share of state AI funding. Part Three proposes four paths forward, including AI audits, stronger contract language, more options for legal recourse, and reprioritizing non-AI government decision-making.

Generating Harms: Generative AI’s Impact & Paths Forward is part of EPIC’s AI & Human Rights Project, which advocates for transparent, equitable, and accountable AI regulations. For more information on EPIC’s report, visit EPIC’s dedicated webpage.