Story perspectives

AI Police Tools Hide Secret Thresholds, Sparking Transparency Concerns

5/12/2026

Summary

Story summary

AI policing tools use invisible confidence thresholds that dictate when an alert triggers police action.
Researchers report that AI systems are deployed in dozens of U.S. cities to score neighborhoods and guide officer deployment.
Vendors or agencies set the thresholds and do not publicly disclose them.
Some police departments follow strict policies linking technology to human discretion, while others lack such guidelines, raising concerns about unchecked use.

History

Historical Context

After the NYPD introduced the CompStat system in 1994, data-driven crime mapping spread to other forces such as the Los Angeles Police Department, reshaping patrol tactics. CompStat’s legacy of secret thresholds parallels today’s AI alerts, while the department already runs a license-plate reader network that has logged more than 2.2 billion reads, raising privacy concerns nationwide.

1 of 2

Story summary

AI policing tools use invisible confidence thresholds that dictate when an alert triggers police action.
Researchers report that AI systems are deployed in dozens of U.S. cities to score neighborhoods and guide officer deployment.
Vendors or agencies set the thresholds and do not publicly disclose them.
Some police departments follow strict policies linking technology to human discretion, while others lack such guidelines, raising concerns about unchecked use.

1 / 2