U.S. AI governance overlooks the sectors experts consider most vulnerable.

Artificial intelligence creates different dangers in different areas of life, but federal rules might not match where the greatest dangers actually are. To see if government attention matches real-world vulnerability, researchers analyzed 684 federal AI governance documents. They looked at how often and how deeply these documents covered 14 sectors and 24 AI risks. Then, they compared these findings to ratings from a study of 272 experts. The researchers found big mismatches. Official documents focus heavily on system security, robustness, and governance, while paying less attention to environmental, socioeconomic, and emerging risks. Furthermore, areas like public administration and national security get high coverage, but healthcare and finance receive comparatively little attention, even though experts consider them highly vulnerable.

Why it matters

Federal agencies decide where to focus AI oversight. This study finds that finance and healthcare—sectors experts say are most vulnerable—receive less governance attention than other areas. That mismatch means regulations may be missing the highest AI risks, potentially leaving the public exposed.

Questions answered