A Police Report Example, Read Three Times

Take any police report example from your last ten cases and you'll find the same anatomy: a narrative, a set of times, an arrest section, a search section, and an evidence log. Most defenders read the narrative once and move on. That's the miss. The case usually lives in the gap between two sources nobody reads side by side, as one defense discovery workflow guide puts it: "Discovery does not compare itself."

Here's the three-pass read that comparison requires.

Pass one: the officer's story. Read the narrative straight through and mark every factual claim. Arrival time, what the officer observed, what the defendant said, where the search happened, what was seized. A report presents the investigating officer's account, and it doesn't include every detail or every perspective. You're mapping the prosecution's theory, not yet testing it.

Pass two: against the timeline. Now put the report's times next to the CAD log, the 911 audio, and the jail calls. The guide's example is worth memorizing: the CAD log says 21:14, the bodycam says 21:31, and the report skips the gap entirely. Seventeen minutes the narrative never explains. That kind of unexplained window is exactly where a motion to suppress lives, because it's where a K-9 sniff, a warrant check run long, or a detention stretching past its justification can hide. Even small timing differences matter when they conflict with other evidence.

Pass three: against the video. Finally, read the narrative claim by claim against the bodycam. Does the "consent" the report describes sound voluntary on the recording? Does the sequence match? Peer-reviewed research from the City University of New York found that across two experiments with more than 1,500 participants, people rated officers as less trustworthy and less accurate when a report was inconsistent with video, and were sometimes more likely to say the officer should be charged. Jurors react to these gaps. So should you, early enough to build cross around them.

The evidence log deserves its own pass-two treatment: every handoff needs a name, a time, and a record. A missing link is a question for cross-examination.

Two caveats keep this honest. First, not every difference in a report changes the case; inconsistency is a lever, not an automatic win, and its significance depends on the record. Second, 2026 reports need one more question on pass one: was any part of this drafted with AI assistance? NACDL now treats AI-generated reports as a new form of evidence defense lawyers must learn to scrutinize, and the DOJ has documented how the report-before-or-after-video question has split policy makers nationally.

When you're running this comparison across hours of bodycam plus a stack of PDFs, tools like Evidence Copilot's AI police report analysis handle the first pass with citations to the exact page or second, while you keep the judgment calls. Check Evidence Copilot pricing if you want to see what the workspace costs per seat.

Why Report-vs-Bodycam Gaps Decide Cases in 2026

A gap between the report and the bodycam can swing a case. Jurors notice, and the research now confirms it.

Two large-scale experiments at CUNY, with more than 1,500 total participants and published in Applied Cognitive Psychology, found that when a police report is described as inconsistent with bodycam video, participants rated the officer as less trustworthy and less accurate, and in some conditions were more likely to say the officer should be charged. Lead researcher Kris-Ann S. Anderson put the takeaway plainly: people overvalue a match between memory and camera, which is why the inconsistency itself becomes the story.

That finding collides with a quieter policy shift that changes what an "initial report" even is. According to a DOJ report issued under Executive Order 14074, the share of agencies allowing officers to view body-worn camera footage before making a statement dropped from roughly 92% to 56% over ten years, based on PERF's 2023 survey of 156 agencies. The emerging model is a two-step process: a perceptual interview from memory first, then a video-informed supplemental report later.

The defense bar is split on whether that model is sound. PERF now recommends the perceptual-interview-then-supplemental sequence. The Major Cities Chiefs Association disagrees, and its 2024 survey found 60% of member agencies still permit pre-statement review. Professor Seth Stoughton, quoted in the same DOJ report, argues that pre-report review "decreases the accuracy of reports, risks diluting the legal standard, and courts public controversy." Arif Alikhan, formerly of LAPD, counters that the practice mirrors memory-inducing procedures used for decades.

For a defender reading a police report example in 2026, the practical questions are:

• Is this the initial perceptual report, or a later supplemental?
• When was it written relative to the officer's video review?
• What changed between versions, and what was left out?
• Does the narrative match the timestamps on the CAD log, bodycam, and 911 audio?

The answer to the last question often decides the motion. As the CUNY team noted, body-worn cameras were sold as a clarity machine, but shaky or obstructed footage frequently produces more uncertainty than certainty, which is exactly where a careful report-vs-video comparison earns its keep.

If you want a faster first pass on the comparison itself, AI police report analysis tools can flag candidate inconsistencies against bodycam and audio, with every answer cited to the exact source moment and marked for attorney review. The Evidence Copilot pricing page lists published tiers with no sales wall, so you can scope it against your caseload before you commit.

Reading the Report for AI Drafting

A police report reads differently once you know an algorithm may have helped write it. In 2026, that question moves from hypothetical to operational, and it belongs in every discovery request you send.

Commercial AI tools that draft police reports have been adopted by some departments, and individual officers use publicly available AI tools on their own, according to a Carnegie Mellon policy paper by Jon M. Peha. NACDL now treats AI-generated reports as a distinct form of evidence that defense lawyers must learn to scrutinize and challenge, with accuracy, bias, and Fourth Amendment exposure as the core issues, per its session materials on decoding AI-generated police reports. EFF has raised the same alarm about police using generative AI to turn body-worn camera recordings into reports without meaningful oversight or controls, as summarized in its commentary on use-based AI regulation.

The scale problem is real. More than 18,000 U.S. law enforcement agencies each make their own AI adoption decision, and most lack the expertise or resources to assess whether any product on the market fits their needs, Peha's paper notes. That means the tool, the training, and the policy can differ from one department to the next, sometimes from one officer to the next.

So ask, in writing, in your first discovery request:

• Was any AI tool used to draft or assist this report?
• Which tool, and which version?
• What source material was fed in (bodycam transcript, prior report, notes, CAD log)?
• Was the officer's perceptual statement preserved separately from any AI-drafted narrative?
• Were draft versions retained, and under what retention schedule?

That last question matters because the anatomy of a report has changed. PERF's 2023 survey of 156 agencies found the share allowing officers to view body-worn camera footage before making a statement after a critical incident dropped from roughly 92% to 56% over ten years, per the DOJ report on BWC policy. The emerging model is a perceptual interview from memory first, then a video-informed supplemental report. If you're reading only one document, you may be reading half the record.

Inconsistency remains your sharpest lever. Two experiments with more than 1,500 total participants found that when a report was described as inconsistent with bodycam video, people rated the officer as less trustworthy and less accurate, and in some cases were more likely to say the officer should be charged, according to CUNY research published in Applied Cognitive Psychology. That's a lever, not an automatic win. As one defense-firm review framework cautions, a difference in a report doesn't automatically change the case, and missing evidence doesn't automatically prove the investigation was improper.

When you run your own first-pass comparison, hold any tool to one test: if it can't show you the page behind each fact, it isn't ready for a suppression motion, as a defense discovery checklist puts it. Evidence Copilot applies that standard to AI police report analysis: every answer cites the exact source moment, output is marked for human review, and originals stay immutable. Published Evidence Copilot pricing starts at $39 per user per month with no sales call.

Running the Comparison: Timeline, Video, and the Page Behind Each Fact

Discovery lands as a box, not a case. The comparison is the work, and the case usually lives in the gap between two sources nobody read side by side.

Start with one timeline. Drop the CAD log, the bodycam clock, the 911 call, and the jail call onto a single line. A CAD log reading 21:14 against a bodycam stamp of 21:31 leaves a 17-minute gap that the narrative never bothers to explain, and that gap is where suppression arguments get built, per this defense discovery workflow guide.

Then run the five-step comparison:

• One timeline: every source on a single clock.
• Report against video: read the narrative while watching the footage.
• Chain of custody: every handoff needs a name, a time, a record.
• Each witness against their first statement.
• Everything referenced but never produced, followed up in writing or by motion to compel.

Send the preservation request with your first discovery request. Video, 911 audio, and jail calls get overwritten on agency retention schedules, and a missing link is a question for cross-examination.

One more question belongs on the list in 2026: was any part of this report AI-drafted? NACDL now treats AI-generated reports as a distinct form of evidence requiring new scrutiny and challenge, and a Carnegie Mellon policy memo notes that more than 18,000 agencies are each deciding on AI adoption without the expertise to assess it.

That is the test for any tool you use on your side of the table. If it cannot show you the page behind each fact, it is not ready for a suppression motion. Grounded AI police report analysis should cite the exact page or second, mark every output for human review, and never reach a legal conclusion. Published Evidence Copilot pricing with no sales wall is how you check that before a client file goes anywhere near it.

A Checklist You Can Run the Day Discovery Arrives

Discovery lands in a dump: a PDF report, some video files with useless names, maybe a 911 audio file. The instinct is to read the report first. Resist it. The case usually lives in the gap between two sources that were never read side by side, which is why a discovery review framework built around comparison beats one built around collection.

Here is the checklist, in order. Run it the same day the files arrive, because some of it is time-sensitive.

• Send a preservation request with your first discovery request. Video, 911 audio, and jail calls can be overwritten under an agency's retention schedule. This is the one item that cannot wait for the second letter.
• Put every source on one timeline. The CAD log says 21:14; the bodycam starts at 21:31. That 17-minute gap the report never explains is a suppression-motion question, and you only find it by reconciling times across documents and media.
• Check the report narrative against the video, second by second. Peer-reviewed research with more than 1,500 participants found that when a report is described as inconsistent with bodycam footage, people rate the officer as less trustworthy and less accurate, and are sometimes more likely to say the officer should be charged. Inconsistency is a lever, though: not every difference changes the case, and the record has to carry the weight.
• Follow the chain of custody. Every handoff needs a name, a time, and a record. A missing link is a question for cross-examination.
• Compare each witness to their first statement, not their latest one. Memory shifts after video review, and the DOJ-commissioned study found the share of agencies letting officers view footage before making a statement dropped from roughly 92% to 56% over ten years. Ask which report you are reading: the initial one or the video-informed supplemental.
• List everything referenced but not produced, and follow up in writing or by motion to compel.

Two 2026-specific items deserve their own line. First, ask whether any part of the report was AI-drafted: NACDL now treats AI-generated reports as a distinct evidence category defense lawyers need to scrutinize, and EFF has flagged generative tools converting bodycam into reports without meaningful oversight. Second, if you use software for the first pass, apply the citation test: if a tool cannot show you the page behind each fact, it is not ready for a suppression motion. That is the standard we built Evidence Copilot around, where AI police report analysis flags report-versus-bodycam conflicts with citations to the exact video moment, and every output is marked pending human review. Published Evidence Copilot pricing starts at $39/user/month on the Solo plan.

If you want to see how this checklist runs against a real discovery dump, book a 20-minute walkthrough at evidencecopilot.com/book, or start on the Solo plan at evidencecopilot.com/signup. No pressure either way; the checklist above works on paper.