Editor’s note: This article is part of the Police1 Leadership Institute, which examines the leadership, policy and operational challenges shaping modern policing. In 2026, the series focuses on artificial intelligence technology and its impact on law enforcement decision-making, risk management and organizational change.
By Chief Robert J. Dowd (Ret.)
At two o’clock in the morning, no patrol officer wants to face a blank screen. The job is done, the arrest is made and the body camera footage is uploading. Somewhere in the building, a sergeant is waiting on paperwork before the tour can finally end. In that exhausted moment, a new temptation enters the squad room: artificial intelligence.
Whether departments have formally addressed it or not, it is happening. The real issue is no longer whether officers will experiment with AI; the issue is whether supervisors are prepared to recognize it, review it and own the results. That is now a core requirement of the job.
| RELATED: Governing AI in policing: What law enforcement leaders need to know
The trade-off: time vs. risk
There is no mystery why this is gaining traction. AI promises the one thing officers never have enough of: time. Tools now exist that can transcribe body-worn camera audio and generate draft narratives in seconds. Some officers are using general AI platforms to organize notes, fix grammar or structure complex reports.
On the surface, this looks like progress. In some ways, it might be. But it also fundamentally changes the risk profile of a police report.
A report is not just paperwork or an administrative assignment; it is a credibility document. It is the foundation for a prosecution, a suppression hearing or a civil case. Every word matters because every word reflects what the officer can testify to under oath. AI introduces a subtle, more insidious problem into that equation: polished inaccuracy. It can produce language that sounds clean and professional but fails to reflect the reality of the encounter.
A report can now be perfectly written, logically structured and still be completely wrong.
The shift from editing to authenticating
For decades, supervisors reviewed reports for completeness, grammar and legal sufficiency. Those standards still matter, but they are no longer enough.
This requires a shift in how we train our sergeants and lieutenants. They are no longer just editors; they are now responsible for evaluating authenticity. They must ask: Does this report reflect what actually happened, or does it reflect what a machine thinks a report should sound like?
Supervisors are no longer just editors; they are now responsible for evaluating authenticity.
Red flags for supervisors
Supervisors need to develop a new skill set for identifying AI influence. Here are three areas where “perfect” writing often fails the reality test.
Sudden stylistic shifts
Writing skills usually improve gradually over a career. When an officer’s writing suddenly becomes flawless, highly polished and stylistically distinct from their past work, it warrants a closer look. Supervisors should have the officer explain the incident in their own words — most of us have done that at a desk more than once. If the verbal explanation lacks the nuance or detail found in the narrative, you have a problem.
Generic conclusions instead of specific facts
AI leans on “cop speak” because that is what it was trained on. It will rely on phrases like “the subject was evasive” or “I feared for my safety.” These are conclusions, not facts. A machine can predict that a nervous suspect avoids eye contact, but it cannot see the specific, minute details that a human officer perceives. If the report lacks observable, articulated details, it should be sent back.
The “smooth” timeline
AI-generated drafts are often suspiciously organized. But real-life scenes are messy, and evidence is rarely linear. If a report reads like a movie script but conflicts with CAD times, body camera footage or physical evidence, the polish is masking a lack of accuracy.
A clean report that contradicts the facts is a liability, not an asset.
Leading through the technology
Consider a sergeant reviewing a vehicle stop report. The narrative is formal and well written, but the description of the suspect’s “nervous behavior” is vague. When the sergeant reviews the body camera, they see the officer actually noted shaking hands and glances toward the center console — details that never made it into the AI-polished draft.
The sergeant sends that report back, not because the grammar was bad but because it lacked the articulation required for court. That must be the new standard of review.
To stay ahead of this, agencies need to move past the “denial” phase and take four practical steps:
- Stop treating this as theoretical. It is not. Officers are already experimenting with these tools. Bring the conversation into the light.
- Create clear policy. If AI-assisted writing is allowed, define the approved tools and what data can or cannot be entered. If it is not allowed, state that clearly and explain why.
- Enforce accountability. The officer must read every line and be prepared to stand behind every word. The supervisor must take the same level of responsibility for approval.
- Train supervisors specifically for this. This is a new tactical skill. Supervisors need to know what AI-influenced writing looks like and how to verify that a report reflects actual observations.
Technology will continue to move forward, and the answer is not to reject it, but to lead through it. Use tools where they help, but never sacrifice the credibility that is the bedrock of our profession.
We must remind every officer of one simple truth: If your name is on the report, you own it. Make sure it reads that way.
About the author
Chief Robert J. Dowd (Ret.) served as the ninth chief of police of the North Bergen (New Jersey) Police Department. He is the chief operating officer of Bernstein Test Prep and an adjunct professor of criminal justice at New Jersey City University. A co-author of “Supervision of Police Personnel,” 10th Edition, he has spent his career training and mentoring law enforcement officers in supervision and leadership with an emphasis on what works in the field.
| NEXT: Why blanket AI bans create more risk than they prevent
