California recently acknowledged the use of several high-risk artificial intelligence systems, underscoring a critical point: AI is increasingly influencing U.S. government decisions impacting millions. According to a Cal Matters report from June 2026, state officials revealed that California agencies were using six high-risk AI systems for tasks ranging from fraud detection to educational oversight. This disclosure contrasts with earlier reports that claimed no such systems were operational.
These AI tools are instrumental in flagging unemployment fraud, monitoring university exams, and spotting suspected AI-generated student work. This situation has sparked concerns about transparency. Critics argue that if governments can’t keep track of their AI systems, the public might lack insight into how these tools shape crucial decisions.
Why does this debate matter? AI is now an integral part of government operations, no longer restricted to experimental trials. Nationwide, states employ algorithms to aid in criminal justice decisions, manage benefits claims, monitor students, and oversee transportation systems. These areas significantly affect people’s rights, finances, and daily lives, and California is not alone in this endeavor.
AI in Criminal Justice
Among the most controversial applications of AI is in criminal justice. California utilizes systems to assess the likelihood of recidivism among incarcerated individuals. States such as Pennsylvania have developed risk-assessment tools to aid judicial decisions.
Algorithms like COMPAS are used in states including Pennsylvania, New York, Wisconsin, and Florida, generating risk scores that influence bail, parole, and sentencing decisions. Critics question whether these systems perpetuate biases, while proponents claim they provide consistency and evidence-based insights.
AI’s Role in Managing Benefits
AI increasingly plays a role in managing benefits programs by identifying fraud and processing large volumes of claims. California uses automated systems for evaluating unemployment claims. Nevada recently adopted an AI-assisted process to draft unemployment appeal rulings for human review.
Automation risks have become evident, as seen in Michigan’s unemployment fraud detection system that inaccurately labeled thousands as fraudsters, prompting a return to human oversight. This case is often cited by advocates seeking stricter safeguards.
AI in Education Systems
AI is rapidly expanding within education systems. California uses tools to detect AI-generated assignments, while Wisconsin has an algorithm-driven early-warning system predicting students at risk of not graduating.
Schools are also trying out AI-powered tutoring systems, personalized learning tools, and administrative software, showcasing the tech’s increasing classroom presence.
Transportation Management with AI
Transportation departments use AI to manage congestion, monitor infrastructure, and make real-time decisions. Texas, Tennessee, and North Carolina have deployed systems analyzing traffic data and adjusting roadway operations based on live conditions.
Supporters believe these systems can alleviate congestion and enhance efficiency. Critics caution that increased automation raises oversight and accountability concerns.
State Leaders in AI Adoption
While AI adoption is widespread, some states like California, Texas, New York, and Virginia are recognized leaders in usage across criminal justice, benefits, transportation, and more.
Other states, including Colorado, Utah, and Virginia, are advancing policy frameworks or pilot programs aimed at AI governance, though details vary.
Overall, AI is no longer experimental in state government. It permeates many systems determining public service delivery and government decisions. Almost every state has tested or implemented AI in fraud detection, education, transportation networks, or administrative roles.
Adoption speeds up as agencies face staffing challenges, increasing workloads, and push for efficiency through automation. The debate now focuses on AI governance. How transparent and accountable should algorithm-influenced decisions be, especially when they affect people’s lives?