- November 22, 2026
- Updated 8:26 am
Michigan Prisons Test New Scoring System for Inmate Security Levels
- 25 Views
- admin
- November 10, 2026
- Technology
In Michigan, incarcerated men participate in employment counseling where security classifications can impact access to job training and educational opportunities. The introduction of a computer-generated scoring system could significantly influence whether someone resides in a low-security dorm or a tightly controlled cell.
The state plans to test a new technology aimed at predicting future prison violence better by approximately 30%. This raises questions about its effectiveness and the accuracy of the technology used.
Does AI Influence Prison Security Levels?
Despite suggestions that artificial intelligence is involved, the nonprofit developing Michigan’s software confirms AI plays no part in calculating security scores. The system follows set rules, with corrections staff making final decisions.
Understanding how this system operates is important. Current methods require employees to manually assess security levels. These levels, ranging from I to V, determine how closely an inmate is supervised and what programs they’re eligible for. Access to education and job training is influenced by these classifications.
What Changes Will The New System Bring?
Starting December 1, selected men’s prisons will pilot a system enabling frequent reviews of an inmate’s security status, rather than being bound to scheduled assessments. This should help staff respond to behavioral changes more promptly.
The software, developed by Recidiviz, evaluates prison records, calculates a score using chosen factors, and can highlight records for further review. Programs that support rehabilitation can also influence someone’s security classification.
Michigan’s Department of Corrections Director, Heidi E. Washington, expresses confidence in the improved safety and refined scoring system, saying it will align security levels with current inmate behaviors and needs.
Explaining the Absence of AI
Recidiviz clarified that AI isn’t used for inmate classification. Instead, a points-based formula, akin to an algorithm, is used. Corrections staff can review and override scores as needed, considering factors like safety or medical requirements.
While AI is employed in other areas of Recidiviz’s platform, it does not influence the security level or release decisions.
Evaluating Safety and Resource Challenges
The touted 30% improvement in violence prediction lacks detailed explanation. Understanding whether the system identifies potential violence or misclassifies individuals is crucial, given the significant impact on an inmate’s daily life.
Correct predictions rely on accurate prison records. Inaccuracies might misinform risk assessments, highlighting the necessity for thorough system evaluations to match housing solutions with security needs.
Michigan plans to expand high-security facilities with 870 new beds as part of the Safe Prisons Initiative. This expansion addresses higher-risk individuals’ housing needs, though some solutions may take over two years to implement.
Union concerns highlight shortages in high-security beds, indicating a need for immediate solutions. Despite technology’s potential, adequate resources and staffing are paramount for effective and safe corrections management.
Looking Ahead
The December 1 pilot could provide insights into the system’s impact on prison safety and fairness for incarcerated individuals. While technology holds promise, its efficacy hinges on integrating human judgment and sufficient corrections infrastructure.
For further discussions or to express your views on technology’s role in safety judgments, share your thoughts with us at CyberGuy.com.