FOX 13 News Utah reporter Mya Constantino says the story starts with a sentence that sounds like it fell out of a cartoon, not a police file.
In a meeting with city leaders, Heber City Police Chief Parker Sever reportedly referenced a report that claimed an officer had “shape-shifted” into a frog.
That line didn’t come from a prankster typing nonsense into a form. Constantino explains it came from AI software the department is testing to help write police reports.
When people saw the claim, it went viral for the obvious reason: police reports are supposed to be boring, factual documents, not fantasy novels.
But Constantino also makes clear the “magical transformation” wasn’t magic at all. It was a machine hearing things it didn’t understand and then confidently writing them down like they were facts.
That’s the part that’s funny at first. And then it gets a little uncomfortable, because police paperwork isn’t just paperwork.
What FOX 13 Says Actually Happened
Constantino explains that the department is using AI tools that generate police reports by analyzing body camera footage.
In the FOX 13 video, Constantino describes the report-writing system as something that can produce a draft quickly, and the officer can click timestamps to check what the camera captured.
The problem, Constantino says, is that the “details aren’t always a hundred percent accurate.”
In the frog situation, Constantino reports that the body cam and the report-writing system picked up audio from a movie playing in the background.
The movie, Constantino says, was “The Princess and the Frog.”
That background audio got folded into the report like it was part of the real-world event, and the result was a sentence that made the officer sound like a character in a fairy tale.
Constantino’s reporting makes the key point: this wasn’t the officer lying. It wasn’t a witness lying. It was a tool trying to turn messy, real audio into clean, official language.
And it got confused.
That is exactly why, according to Constantino, the department emphasizes that officers still review the drafts before anything becomes final.
Riding Along To Watch The AI Work
Constantino doesn’t just describe the system from a distance. She shows herself stepping into a police cruiser to watch how the process works.
In her FOX 13 report, Constantino says she rode along with Sergeant Rick Keel for a demonstration.
During the staged stop, Keel gives the kind of standard opening you’d expect, introducing himself and explaining why he’s stopping someone.
Constantino adds a little human moment in the middle of the demo, asking questions that regular viewers would ask, including whether officers are watching streaming content while driving.
After the camera recording is saved, Constantino explains the workflow: dock the camera, and the system produces a report draft.
Back at the police department, Constantino says the AI generated a report with timestamps, and the draft was mostly normal.
But she points out it still needed corrections, even in a controlled demo where everyone was trying to keep it simple.
In the video, Constantino describes one example where the AI misread the scene. Something parked on the side of the road got turned into something traveling east, because the system apparently interpreted the context wrong.
That kind of error doesn’t go viral like the frog line, but it’s actually the more serious type, because it looks believable.
The Big Selling Point: Time Saved
Constantino reports that a major reason the department is testing AI is time.
She says Sergeant Keel told her that writing a report the old way often takes one to two hours.
And Keel says the tool is now saving him six to eight hours every week.
That is not a small perk. That’s real time – time that can go toward patrol, calls, or just not staying late to finish paperwork.
Constantino also reports that the trial run was scheduled to wrap up in January, and department officials were leaning toward continuing with AI report tools going forward, even if they still need to decide which system fits best.
Here’s where the story gets tricky. Saving time is great, and everyone wants officers doing less typing and more real police work.
But when speed is the prize, “review it carefully” can start turning into “skim it quickly,” especially when supervisors are pressured to do more with less.
That pressure doesn’t have to be evil to be dangerous. It just has to exist.
Steve Lehto’s Warning: The Courtroom Is Where This Blows Up
Attorney and YouTuber Steve Lehto, on Lehto’s Law, says this story spread fast because it hits a topic he talks about a lot: AI’s “foibles,” and how those flaws tend to show up in the news.
Lehto highlights the core issue the frog line reveals: the body cam might capture extra background noise, and AI might treat it like meaningful speech or facts.
He says sometimes AI will even “hallucinate,” producing statements that don’t match anything you can find on the actual footage.
Lehto argues the real nightmare isn’t the internet laughing about a frog. It’s what happens later when a report goes into court.
He walks through a scenario where a defense attorney cross-examines an officer using a report that contains a clear AI-generated error.
Lehto’s point is blunt: if you’ve got a report with something obviously wrong in it, the defense doesn’t even need to be clever. The jury starts laughing, confidence drops, and the officer’s credibility takes a hit.
Lehto stresses a simple fact that’s easy to forget: even if a machine drafted the words, the officer signs the report.
And once it’s signed, it’s not “the computer’s mistake.” It’s the report the officer swore was accurate.
Lehto also raises a practical concern that feels inevitable. If the department adopts this to save time, leadership may push officers to tighten their review process into something faster, like “just skim it and tidy it up.”
That’s how mistakes slip through – not because people are lazy, but because the system is designed to reward speed.
The Frog Joke Is Funny, But The Lesson Isn’t
Constantino’s FOX 13 piece treats the frog report as a strange, funny moment that exposed a simple cause: background audio.
Lehto agrees the cause is simple, but he says the bigger lesson is harsh: AI can write sentences that look official while having no built-in understanding of true versus false.
Lehto’s warning is basically this: AI can assemble words into something that sounds authoritative, but it does not “know” reality the way a human witness does.
So if the AI hears a movie clip, or picks up a random phrase, it may drop it into the report without any common sense filter.
And police reports are not a place where you can shrug and say, “Well, that’s weird.”
They are used to justify arrests, guide charging decisions, shape plea negotiations, and influence outcomes. Even small inaccuracies can snowball, because the next person assumes the report was checked.
The frog line grabbed attention because it’s absurd. But absurdity is a gift, because it’s obvious.
The scarier errors are the quiet ones – wrong direction of travel, wrong sequence of events, wrong tone, wrong “admission” that wasn’t really spoken – things that don’t look crazy until someone compares them closely to video.
A Reasonable Use, With A Dangerous Edge
Constantino and Lehto both, in their own ways, point to the same middle ground: AI might be useful as a draft tool, but only if humans treat it like a rough draft, not like truth.
Constantino reports that officers review the drafts, and that the frog situation taught them the importance of correcting AI-generated reports.
Lehto’s problem is that “importance” doesn’t always survive contact with real life. Not when calls stack up, staffing runs thin, and the entire reason you bought the software was to save hours.
This story reads like a comedy headline, but it’s really a warning label for modern paperwork.
If police departments are going to rely on AI to speed up report writing, there has to be a culture that treats the AI output as suspect by default.
Because the second it becomes “good enough,” the system starts producing official-looking mistakes with a badge on them.
And once that happens, the frog won’t be the most embarrassing part.
It’ll be the moment everyone realizes the report was treated like a fact document, even when it was partly a guess.
Gary’s love for adventure and preparedness stems from his background as a former Army medic. Having served in remote locations around the world, he knows the importance of being ready for any situation, whether in the wilderness or urban environments. Gary’s practical medical expertise blends with his passion for outdoor survival, making him an expert in both emergency medical care and rugged, off-the-grid living. He writes to equip readers with the skills needed to stay safe and resilient in any scenario.
