Artificial Intelligence in Policing: Predictive Policing, Civil Rights, and the Future of Crime Analysis

Law enforcement is adopting AI fast—but not always wisely. Here’s what history tells us.

Oct 07, 2025

AI in Policing: Buzzword vs. Reality

Artificial Intelligence (AI) is everywhere. Every product offered today has some AI component. While AI is constantly in conversations, most anecdotes involve the relief of simple tasks. Like the advent of the accessible Internet thirty years ago, today it has more power as a buzzword than as an actual solution in most cases.

In the application of policing, there are concerns outside of law enforcement regarding both privacy and civil rights issues. The National Association for the Advancement of Colored People (NAACP) is very concerned, from an AI perspective, with predictive policing. In their Artificial Intelligence in Predictive Policing Issue Brief, they state:

“Jurisdictions who use this tool argue it enhances public safety, but in reality, there is growing evidence that AI-driven predictive policing perpetuates racial bias, violates privacy rights, and undermines public trust in law enforcement. The data used to make decisions around predictive policing comes from compiling and analyzing historical criminal data and police activity. Relying on historical criminal data to make policing decisions is inherently biased, as data shows that the Black community is disproportionately negatively impacted in the criminal justice system due to targeted over-policing and discriminatory criminal laws.”

While there may be some legitimate concerns about AI, predictive policing predates it by 80 years taking root in the 1930s and going through many states and forms prior to it being enhanced with AI. Crime analysis as a discipline relies upon historical data to detect patterns and develop strategies to address crime in all neighborhoods. The AI layer of predictive policing did not emerge until the mid-2010s. The NAACP is not as opposed to AI, which ironically would automate processes of analysis, free of human bias, as they are to crime analysis in general.

The National Conference of State Legislators published a report on AI in law enforcement in February 2025 that listed the areas of concern as facial recognition, drone technology and automated license plate readers (ALPRs.) Again, these are all existing technologies that predate AI, and now some are AI enhanced – are addressed in the report by their core function, not the addition of AI. At least 18 states have pushed legislation on both the use of facial recognition, especially via body worn cameras, and ALPRs.

The Real Impact: Reporting, Analysts, and the Future of Crimefighting

Recently I was invited to a webinar on Artificial Intelligence in Policing hosted by Capt. Mike Schentrup (ret.) of Advanced Police Concepts, one of the premier law enforcement training companies. It was an informative event as the hosts focused on AI-assisted police incident report writing. This was interesting with a deep dive that included a supervisor’s experience as well as an end user’s perspective. I learned from this webinar that the vendor includes outrageous false statements to prompt officers to proofread their AI-generated reports before submitting. Vendors point to statistics claiming that officers spend 25-40% of their shift time on completing reports with the promise their product can reclaim this time so cops can spend more time fighting crime.

I am all for police officers having the best of technology at their fingertips. There are a few reasons why the adoption of AI assisted reporting is not the best tactic. My first impression is that it will make a lazy cop lazier. Three decades ago, when several major jurisdictions discontinued police responses to several ‘over-with’ crime categories due to fiscal realities, we learned that if law enforcement didn’t respond and determine if there was a scene to process, there was no hope of either solving the cases or linking them to other solved cases. It doesn’t take long for citizens to get the hint, especially with self-reporting or online filing of cases, that the cops just don’t care. Unless there is either identity theft or a loss covered by insurance, there is little incentive for victims to report their crimes.

This approach is in contradiction to some of the foundational studies that determined factors including speed of response were not related to the solvability of crimes. We may have slowed our response, but officers still showed up. It is continually proven that investigating cases solves crime just as enforcing so-called low-level crime functions to control serious crime. I can’t speak for every agency, but I know in some larger departments, the patrol ecosystem depends on the call takers to free up the officers who want to engage in self-initiated crimefighting. Given that dynamic, although the vendors statistics are likely fairly accurate, freeing up more time in report writing via AI is unlikely to result in either more thorough investigations or more proactive patrol. In fact, the result I predict will be less solvable cases.

Another hazard facing policing through this AI revolution will be the inclination to either eliminate or just not replace civilian analysts. This would be a mistake in the pattern of the “Doorman Fallacy.” The story goes: the superintendent of an apartment building was persuaded to dismiss a doorman who didn’t appear to contribute much, but after he was gone, from leaves in the lobby, a dirty main door, to litter on the sidewalk and nuisance people loitering at the entrance, his absence made a huge difference. We will have to remember that the ability to tell a GPT to put dots on a map doesn’t make an officer an analyst. It is easy to see many of these positions be deemed as obsolete, and many legitimately will be, but the rivers of raw data will continue to grow and expand, and few will have the background and education to navigate the next chapter of crime and intelligence analysis.

Leading Through the AI Shift

Will AI benefit policing? It undoubtably will. As a profession, AI needs to be adopted with the vision and clarity that it is a technology that solves problems. It will have a role in crimefighting, if we give it tasks that demand reasoning. A great deal will have to do with how we train our staff to properly and responsibly exploit the capabilities.

We also need to be good students of history. For many, the advent of the Information Superhighway around 1997, plateaued with the AOL ‘You’ve got mail!’ alert and a dancing baby screensaver. The integration of technology over the last quarter century from 800 MHz radios with 160 talk-groups to Rapid DNA and the array of technology previously mentioned should have caused a tremendous reduction in crime, but it hasn’t. It appears that only one thing works everywhere it is tried: getting back to the basics of policing!

Artificial Intelligence in Policing: Predictive Policing, Civil Rights, and the Future of Crime Analysis

Law enforcement is adopting AI fast—but not always wisely. Here’s what history tells us.

AI in Policing: Buzzword vs. Reality

The Real Impact: Reporting, Analysts, and the Future of Crimefighting

Leading Through the AI Shift

Discussion about this post