On 14 April, the Trump administration quietly acknowledged the widespread use of AI to automate government processes. The Office of Management and Budget (OMB) disclosed 3,611 active or planned AI use cases across the federal government, a staggering 70% increase from the list published in the final year of the Biden administration. This expansion includes many plans to hand over sensitive governmental functions to AI, raising alarms about transparency and public involvement.
Disturbing Examples of AI Deployment
Scanning the list, many readers may find cause for alarm. The Health and Human Services (HHS) office for children and families hired Palantir, a company known for its work with the military and intelligence agencies, to scan grant applications for ideological alignment. The Federal Bureau of Prisons is developing an AI to assess misconduct potential in new inmates, potentially routing them into high-security confinement before any wrongdoing. The Department of Veterans Affairs is creating an AI to monitor crisis line calls and assess suicide risk using external databases. The Department of Energy is testing AI to control nuclear reactors autonomously. Meanwhile, the State Department ended a program using AI to forecast mass civilian killings, intended for conflict prevention.
Potential for Responsible Use
While these uses sound alarming, any could be implemented responsibly. The HHS system might enforce policy alignment, but the concern is more about the policy itself. The VA's suicide prevention AI could save lives if validated. Predictive methods for prisoner classification are decades old, and autonomous nuclear reactor control is well studied. However, the inventory lacks crucial details: descriptions are typically a sentence long, and public consultation is minimal. Only one example (the DoJ) proposes public involvement, as others are not classified as 'high impact.'
Need for Transparency and Public Input
AI disclosure could build trust if paired with meaningful public consultation. Washington DC and California are engaging the public on AI use in government. France's Digital Republic Act sets a gold standard, requiring algorithms used in administrative decisions to be subject to public records, human appeal, and mandatory notification. Canada's 2025 AI registry includes a transparent risk-scoring and impact assessment process, though it could be improved with public comment periods. The US should adopt algorithmic impact assessments and public comment processes to ensure safe, trusted, equitable AI transformation.
