Skip to content
logo The magazine for digital lifestyle and entertainment
A brand of Bild
Artificial intelligence News Security All topics
Pilot Project in U.S. Prisons

Can AI Predict Violent Acts?

A person is standing with their back to the camera, wearing handcuffs behind their back. They are dressed in a blue sweater and blue jeans. The background is blurred, showing vertical gray structures.
AI is intended to assist in the placement of inmates and the assessment of violence risk. Photo: Getty Images
Prefer TECHBOOK on Google Prefer on Google Share article

November 7, 2026, 7:14 am | Read time: 2 minutes

Artificial intelligence is set to help assess the risk of violence among inmates in U.S. prisons. The state of Michigan plans to use AI-based software for this purpose. It will evaluate prisoners based on their potential risk and assign them to one of five security levels.

The corrections department expects this to provide an assessment about 30 percent more accurate than the current method. However, experts warn of potential errors and biases in the underlying data. These could have significant impacts on prisoners who are incorrectly classified.

Software to Help Determine Security Level

Currently, a panel of prison staff decides the classification of inmates. In the future, software from the nonprofit technology organization Recidiviz will support this process. The organization originated from a data project at Google and develops digital solutions for the U.S. correctional system.

According to the corrections department, testing begins in December 2026 in several men’s prisons. The AI will help organize inmate housing and more accurately assess the risk of potential violent acts.

More on the topic

Faulty Data Could Have Consequences for Inmates

The classification directly affects prison life. The security level determines which other inmates a prisoner is housed with. It also dictates which programs and support services are available to them.

The reliability of the software’s judgment depends significantly on the data it uses. Futurism points to a ProPublica investigation of a similar risk assessment program in Louisiana. The data used for classification were racially biased and disadvantaged social backgrounds, leading to unjustifiably higher risk classifications for some inmates.

Also of interest: AI Tax in Germany? This Proposal Now Comes from the CDU

Berlin Also Tests AI for Violence Detection

Not only in the U.S. is artificial intelligence used to detect potential violent acts. Germany also has a corresponding test project. At Kottbusser Tor in Berlin, AI analyzes video footage. The system is designed to identify possible violent situations in this way.

This article is a machine translation of the original German version of TECHBOOK and has been reviewed for accuracy and quality by a native speaker. For feedback, please contact us at info@techbook.de.

You have successfully withdrawn your consent to the processing of personal data through tracking and advertising when using this website. You can now consent to data processing again or object to legitimate interests.