- November 22, 2026
- Updated 9:01 am
Michigan Introduces New Prisoner Classification System
Michigan is set to trial a new prisoner classification system starting December 1. This system aims to enhance the prediction of potential future violence by about 30 percent, according to the Michigan Department of Corrections (MDOC). The move comes in response to a significant rise in prison violence between 2019 and 2024, with assaults on prison staff increasing by 77 percent and incidents among incarcerated individuals rising by 54 percent.
The trial will be conducted in selected men’s facilities. The classifications determined by this system will influence prisoner movement, supervision, and access to certain programs and activities. Recidiviz, a nonprofit technology organization collaborating on this project, clarified that the system does not utilize artificial intelligence for decision-making regarding security levels or releases.
Revamping the Approach
Currently, Michigan employs four security levels: I, II, IV, and V, skipping Level III. These classifications significantly impact prisoner experiences, determining their freedom of movement, supervision intensity, and access to opportunities like vocational and educational programs.
With the new system, classification reviews will use real-time data analyzed continuously, which is expected to improve the accuracy of violence prediction by approximately 30 percent. However, how this improvement figure is calculated has not been disclosed in public announcements.
Objective Behind the New System
MDOC Director Heidi E. Washington emphasized the objective nature of the pilot system, which uses evidence-based historical data to enhance facility security. Placing individuals at security levels that reflect their current behavior and needs protects both incarcerated individuals and corrections staff.
The system’s pilot in December will assess its effectiveness before a potential wider implementation throughout the state.
Concerns and Transparency
Algorithmic assessments in criminal justice have previously raised concerns about transparency and potential biases. For example, a ProPublica investigation into a different system in Louisiana highlighted disparities in how data affects decision-making. Although Michigan’s system focuses on security classifications rather than reoffending, questions remain about data sources, predictive accuracy, staff intervention capabilities, and bias detection safeguards. These factors are crucial as Michigan evaluates the system’s long-term integration.
Recent Posts
- The Persistent Tension Between American Profanity and Censorship
- Tommy Lee Jones on Life in Texas and His Passion for Polo
- Challenges in California’s Film Industry as Production Shifts Out of State
- Outrage Follows Mike Florio’s Holocaust Joke on NFL Show
- Tyler Lundblade’s Impact On and Off the Court