I really wish policing would take more inspiration from aviation on a different avenue for police accountability.
The NTSB exists not to blame pilots (though they sometimes do), but to make air travel safer and prevent future plane crashes. In the business of preventing disaster in safety-critical industry, if you chalk something up to human error or call it a tragic accident, you guarantee that it will happen again. Finding that everyone did everything by the book means the book needs to be rewritten because the book that exists today contains a recipe for plane crashes.
I wish police would treat use of force incidents the same way. The investigations after police use of force ask whether the officer violated the law or department policy. Like most law enforcement and judicial work, the exercise focuses on identifying, trying, and punishing guilty parties. If there is no guilty party, the process can produce no change. I would like to see more investigations into police use of force that focus on improving safety outcomes instead.
What about less? Take away guns and reach of the cops and politicians?
Accountability by making 900k cops across all levels of government stripped of power and made normal people? Same for the 600k politicians coast to coast. Screw their story mode mental illness.
Make everyone busy generalizing logistics process to serve biology and stop with story mode hustling memes about fiat (vacuous proclamations) valuations using jargon from the 1800s?
Roughly 1.5 million pols and cops have 10s of millions wrapped around their finger. With urbanization the best part is a bunch of them live just a few miles from any given large urban area full of people being screwed by them.
The time for demanding meager reforms from 60+ year olds who have no skin in our future is long gone.
Except the low level gossipy kind like “so n so cheated”. Statistical analysis of death trends suggest we kill each other on Main Street over such gossip at the same rate humans did centuries ago. It’s those moments of nation state fueled atrocity and imperialism when human death spikes. Seem clear in the streets most adults just don’t go on murderous rampage.
Yes. It doesn’t matter exactly how each word in the police report was entered. All that matters is the officer signed off on it. They should be personally & totally responsible for the contents of the report. I don’t care if they use generative AI, speech to text, Dvorak touch typing, QWERTY hunt-and-peck or anything else. An officer must read the final report and sign to assert its accuracy.
If police reports are low quality, it’s an officer performance problem. Obviously performance management in public safety is exceptionally challenging, but that’s the problem domain that matters. You cannot solve law enforcement accountability by tweaking your AI User Interface.
That said, this seems like a missed opportunity to use technology to increase accountability. If you’re running speech to text on body cam footage, great! Everyone involved in the conversation should get a copy of the transcript. There should be a straightforward way to challenge STT errors.
Again though, it’s the same deal as the body cam footage itself. Always-on body cams with default public access are one thing, officer-managed, sue-to-review is quite another. The crucial issues are political, not technical.
It should matter which parts were written by AI or by officer. Once the officer signs off on the report, they take full responsibility for the content.
Yeah; Law Enforcement & the Judiciary is going to be an early flashpoint in the "a computer must never make a management decision" conflict. IMO: Its actually really important that these systems do not explicitly mark content as AI generated versus not, because if the operator is going to be held responsible for the final content, we can't allow repudiation of that content through an argument "well, that part was AI generated". Even if it isn't written with their voice; it should be reviewed and accepted by them, and at that point it doesn't matter if the AI writes it.
This is a contrived metaphor, but imagine some case report makes its way to a judge, and its missing some significant details about the case. But, its hand-written, and the officer argues that hand-writing is physically harder than typing, so of course hand-written reports wont be as comprehensive as typed ones. That argument is insane, partly because its an imperfect metaphor, but the line of logic is there.
There is no accountability behind that responsibility.
Cops are not held to account for lying now. Even when they are caught, 99% of the time the worst consequence for them is their testimony is ignored in court. They don't face professional repercussions in practice.
So even if officers are responsible for their reports they will still take the easy route and sign off on AI garbage. There is no downside, and it helps them meet pressure from their bosses and avoid the part of the job they hate the most.
Having picked a habit of watching propable cause proceedings on YouTube, I wonder if this is simply result of real reports that AI was trained on being purposefully obtuse and laconic to give prosecutors a wiggle space in the court room?
The prosecutors are given the most absolute trash reports to work with. "Failure to ID, after a traffic stop." "What was the stop for?" "It doesn't say." "So no PC for the stop."
"A caller and said she thought someone was stealing their neighbor's U Haul. A man was observed walking on that street and taken into custody for ..." "For what? Walking while black?"
But no sympathy for the prosecutors either. Garbage reports, but they obviously don't read them pre-hearing, and have plainly become accustomed to judges rubber stamping their PC hearings.
I do like that he doesn't go 'lightly' with the defendants. "You got off lucky this time. You know it, I know it. Do better or it might not go the same next time", and when there is PC or other such, he doesn't put up with any bullshit either.
What Axon's product should be: Define "best" police report, and assist the officer to write that.
What it is: Axon makes whatever police departments ask for.
It doesn't have to be a big conspiracy. It's not incompetence either. Hanlon's Razer should really be, "Never attribute to malice that which is adequately explained by" the enterprise sales pipeline.
Enterprise sales is why we are talking about Axon and not far older, detailed, thoughtful efforts from all sorts of other organizations.
our officers don’t have time to comb through every transcript, fixing it for privacy, empathy, and all that. But keeping the transcripts is still a big win: more info in police records can make police officers more data-driven :)
I think such a tool could be useful for ensuring all the facts get included, but I hate the idea that some departments could start highering illiterate officers if this tech goes far enough.
> So we don’t store the original draft and that’s by design and that’s really because the last thing we want to do is create more disclosure headaches for our customers and our attorney’s offices
You have to wonder if this will stand up in court. I hope not.
AI has a great opportunity to take processes that contain hidden bias and make them more legible and therefore amenable to fixing.
But it also has the opportunity to do the opposite, and we should be cautious to make sure guardrails are in place when putting this tech into life-and-death systems.
“Stamp this LLM text in a hurry” is an invitation for whatever errors and biases are baked into the system to be propagated. You need provenance and measurement of LLM outputs.
I wondering how much this even matters in the age of everything being recorded.
If they are using axon body cameras and vehicle cameras, then usually the entire interaction is recorded, often from multiple officers.
I cannot imagine a defense so incompetent that they rely on the police report rather than watching the entire body cam footage and doing their own assessment.
Even if the cops are doing something sketchy (like turning off their camera) then it's not like the police report would be any more trustworthy.
The current administration has already removed the requirement for federal police forces to wear body cameras. As well as made statements (but little action so far) to federalize the police force to be under the jurisdiction of the DOJ. Everything being recorded may not be the case very soon.
Sorry, I’d get sources but I just woke up, I’ll edit this later with them.
It goes a lot deeper than this, the real world isn't as simple as 'objective truth' and much of the law relies on interpreting the facts we all seek. This is where this technology fails, it normalizes nudging the margins to include a framing of what happened (including that video) using particular and precise language. That language influences court decisions.
For example, the phrase 'furtive movements' seems really anochronistic. Is that a phrase you use? cops use in their day to day life? But it constantly shows up in police reports. Why? The courts have said that 'furtive' movements are suspicious enough to trigger probable cause - which justifies a search. So now, cops every where write that they observe movements that are furtive. Is what your attorney viewed furtive? where they normal movements? were they suspicious? The cop described them as furtive though and we defer to cops, in part because they speak the language of the courts, and now your arrest is valid and that search is valid and whatever is recovered is valid - because a court said movements need to be furtive and you sneezed and a cop described that as furtive even though he had already decided to do the search before he got out of his car.
The only way our system works is if at every level every participant (people, jurors, judges, politicians) distrust the words of police - especially when they habitually use the language of the law to justify their actions. What this tool does is quite the opposite, it will statistically normalize the words police use to describe every interaction in language that is meant to persuade and influence courts now and over time to defer to police.
I was thinking the same thing. If the AI report depends on the raw audio, then it should be preserved and the defense should compare that to the final police report. Having the edit history would be useful for improving the software and analyzing the officer's motivations, but ultimately we're not in a worse situation than before.
I'd predict the synthesis of the AI transcript and the police officer's memory will be more accurate than just the police officer alone. Would be nice if there's an independent study.
There are very incompetent public defenders, if we attribute to incompetence instead of malice, AI isn't changing that.
Even in jurisdictions that require recordings at all times, there are times when the police are required by law to switch them off (entering certain non-public spaces etc), so there can always be gaps that are legal, never mind illegal.
I cannot imagine a defense so incompetent that they rely on the police report rather than watching the entire body cam footage and doing their own assessment.
Not incompetent at all. Police officers often turn in multiple reports following an incident and arrest. Sometimes they contradict each other, and it would be a foolish defense attorney who did not explore those contradictions.
Body cameras capture a lot less than you'd think. Lets just take one example: you can't see the cop at all.
Imagine a cop says "stand over there" but a suspect doesn't know where they mean and doesn't move. Axon might well add "I pointed to the kerb" on the police report, which the police officer rubber stamps. Now the system has invented a clear non-complying suspect with nothing but their word to refute it.
Similarly shaky camera, limited field of view, poor audio, periods when cameras are off all add opportunities for camera footage to fail to prove what happened. Police reports have a lot of weight in court, they are treated as true by default.
This article is about Axon purposely working to make audits harder and police less accountable. Is it so hard to imagine a future feature that summarizes the body cam footage then erases it?
> sign an acknowledgement that the report was generated using Draft One and that they have reviewed the report and made necessary edits to ensure it is consistent with the officer’s recollection.
We already know that police officers are not more reliable than the general public as eye witnesses and that eye witness reports are generally very unreliable as they are very susceptible to prompting bias. This seems like leaning in to prompt bias. The AI is now prompting the human rather than the other way around. This is perverse.
This requires audio to work and appears to create more transparency. You can request the audio recording to verify accuracy. This will happen as a routine procedure from defense attorneys. Any problems with the technology would be discovered quickly and if the officer didn’t do their job of correcting the errors before the report is generated they would be torn apart.
You can verify the accuracy of a transcript, but a police report is much much more than a transcript. It includes off-camera action, smells, states of mind, impressions, context, facial expressions, etc. If Axon hallucinates that stuff, there's no way you can use video footage to check the accuracy or refute the statement.
When EU introduced its AI Act there was much gnashing of teeth here at HN over "stifling of innovation" and "getting left behind in technological backwater".
EU AI Act specifically calls out and forbids such applications. Of course, the state will do what the state will do, but there's an actual obstacle enshrined in law.
A lot of people worry about a Terminator style AI apocalypse. I don’t.
I worry that we’ve already created the AI apocalypse and that this is what it looks like, along with extremist magnification on social media.
I trust AI to be what is is- essentially a lot of math that classisfies and predicts stuff, usually words, that the prediction can be used generatively, and the classification stuff can be used to identify stuff in various media.
What I don’t trust is that people will use it responsibly. Hell, I don’t, when I’m vibe coding, but that’s on me.
People are venal and self absorbed and busy and lazy and all the other traits that lead to not using AI responsibly. And businesses are amoral (not immoral) and want the shortest path to revenue, with the least friction.
So of course police officers who want to be on patrol and did not sign up to spend countless hours on reports, are going to let the AI write it and call it good without proofreading.
We could pass a lot of laws trying to specify products that force police to act reliably, or we could maybe just pass a law that says that AI cannot be used to write police reports, but that clearly labeled AI generated transcriptions and summaries may be attached unedited to police reports, if and only if the original recordings are also preserved as evidence.
And police departments that keep body camera and car camera footage might be ease up on the report writing and only require officers to annotate it with their impressions, but otherwise let the record speak for itself.
Creative lawyers will be all over this. First you get the officer to testify that AI helped write the report, then you call the AI as a witness. When the judge tosses that, you start issuing subpoenas to everyone you can find at OpenAI and Axon.
As others point out, the actual bodycam footage will be definitively probative for the events it records. But there are plenty of cases where the report itself leads to later actions that may be tortious or criminal, and finding out who's to blame for the exact wording used is highly relevant.
Example: AI incorrectly reports that during A's arrest, A made incriminating allegations about B. Based on the report, the police get a warrant and search B's house. When it turns out B is innocent, B sues the department, and when the report turns up during discovery, we're off to the circus.
32 comments
[ 3.1 ms ] story [ 59.4 ms ] thread[1] yes they are: https://www.google.com/search?q=are+pilots+accountable+for+u...
The NTSB exists not to blame pilots (though they sometimes do), but to make air travel safer and prevent future plane crashes. In the business of preventing disaster in safety-critical industry, if you chalk something up to human error or call it a tragic accident, you guarantee that it will happen again. Finding that everyone did everything by the book means the book needs to be rewritten because the book that exists today contains a recipe for plane crashes.
I wish police would treat use of force incidents the same way. The investigations after police use of force ask whether the officer violated the law or department policy. Like most law enforcement and judicial work, the exercise focuses on identifying, trying, and punishing guilty parties. If there is no guilty party, the process can produce no change. I would like to see more investigations into police use of force that focus on improving safety outcomes instead.
What about less? Take away guns and reach of the cops and politicians?
Accountability by making 900k cops across all levels of government stripped of power and made normal people? Same for the 600k politicians coast to coast. Screw their story mode mental illness.
Make everyone busy generalizing logistics process to serve biology and stop with story mode hustling memes about fiat (vacuous proclamations) valuations using jargon from the 1800s?
Roughly 1.5 million pols and cops have 10s of millions wrapped around their finger. With urbanization the best part is a bunch of them live just a few miles from any given large urban area full of people being screwed by them.
The time for demanding meager reforms from 60+ year olds who have no skin in our future is long gone.
Skip the guns and go the route of making everyone a normie civil servant and no one has leverage https://aeon.co/essays/game-theory-s-cure-for-corruption-mak...
Except the low level gossipy kind like “so n so cheated”. Statistical analysis of death trends suggest we kill each other on Main Street over such gossip at the same rate humans did centuries ago. It’s those moments of nation state fueled atrocity and imperialism when human death spikes. Seem clear in the streets most adults just don’t go on murderous rampage.
If police reports are low quality, it’s an officer performance problem. Obviously performance management in public safety is exceptionally challenging, but that’s the problem domain that matters. You cannot solve law enforcement accountability by tweaking your AI User Interface.
That said, this seems like a missed opportunity to use technology to increase accountability. If you’re running speech to text on body cam footage, great! Everyone involved in the conversation should get a copy of the transcript. There should be a straightforward way to challenge STT errors.
Again though, it’s the same deal as the body cam footage itself. Always-on body cams with default public access are one thing, officer-managed, sue-to-review is quite another. The crucial issues are political, not technical.
This is a contrived metaphor, but imagine some case report makes its way to a judge, and its missing some significant details about the case. But, its hand-written, and the officer argues that hand-writing is physically harder than typing, so of course hand-written reports wont be as comprehensive as typed ones. That argument is insane, partly because its an imperfect metaphor, but the line of logic is there.
Cops are not held to account for lying now. Even when they are caught, 99% of the time the worst consequence for them is their testimony is ignored in court. They don't face professional repercussions in practice.
So even if officers are responsible for their reports they will still take the easy route and sign off on AI garbage. There is no downside, and it helps them meet pressure from their bosses and avoid the part of the job they hate the most.
I do enjoy seeing those (well, I shouldn't).
The prosecutors are given the most absolute trash reports to work with. "Failure to ID, after a traffic stop." "What was the stop for?" "It doesn't say." "So no PC for the stop."
"A caller and said she thought someone was stealing their neighbor's U Haul. A man was observed walking on that street and taken into custody for ..." "For what? Walking while black?"
But no sympathy for the prosecutors either. Garbage reports, but they obviously don't read them pre-hearing, and have plainly become accustomed to judges rubber stamping their PC hearings.
I do like that he doesn't go 'lightly' with the defendants. "You got off lucky this time. You know it, I know it. Do better or it might not go the same next time", and when there is PC or other such, he doesn't put up with any bullshit either.
More judges like him are needed.
What Axon's product should be: Define "best" police report, and assist the officer to write that.
What it is: Axon makes whatever police departments ask for.
It doesn't have to be a big conspiracy. It's not incompetence either. Hanlon's Razer should really be, "Never attribute to malice that which is adequately explained by" the enterprise sales pipeline.
Enterprise sales is why we are talking about Axon and not far older, detailed, thoughtful efforts from all sorts of other organizations.
You have to wonder if this will stand up in court. I hope not.
AI has a great opportunity to take processes that contain hidden bias and make them more legible and therefore amenable to fixing.
But it also has the opportunity to do the opposite, and we should be cautious to make sure guardrails are in place when putting this tech into life-and-death systems.
“Stamp this LLM text in a hurry” is an invitation for whatever errors and biases are baked into the system to be propagated. You need provenance and measurement of LLM outputs.
If they are using axon body cameras and vehicle cameras, then usually the entire interaction is recorded, often from multiple officers.
I cannot imagine a defense so incompetent that they rely on the police report rather than watching the entire body cam footage and doing their own assessment.
Even if the cops are doing something sketchy (like turning off their camera) then it's not like the police report would be any more trustworthy.
For example, the phrase 'furtive movements' seems really anochronistic. Is that a phrase you use? cops use in their day to day life? But it constantly shows up in police reports. Why? The courts have said that 'furtive' movements are suspicious enough to trigger probable cause - which justifies a search. So now, cops every where write that they observe movements that are furtive. Is what your attorney viewed furtive? where they normal movements? were they suspicious? The cop described them as furtive though and we defer to cops, in part because they speak the language of the courts, and now your arrest is valid and that search is valid and whatever is recovered is valid - because a court said movements need to be furtive and you sneezed and a cop described that as furtive even though he had already decided to do the search before he got out of his car.
The only way our system works is if at every level every participant (people, jurors, judges, politicians) distrust the words of police - especially when they habitually use the language of the law to justify their actions. What this tool does is quite the opposite, it will statistically normalize the words police use to describe every interaction in language that is meant to persuade and influence courts now and over time to defer to police.
https://www.bjjohnsonlaw.com/furtive-movements-and-fourth-am...
https://www.californialawreview.org/print/whack-a-mole-sus
I'd predict the synthesis of the AI transcript and the police officer's memory will be more accurate than just the police officer alone. Would be nice if there's an independent study.
There are very incompetent public defenders, if we attribute to incompetence instead of malice, AI isn't changing that.
Not incompetent at all. Police officers often turn in multiple reports following an incident and arrest. Sometimes they contradict each other, and it would be a foolish defense attorney who did not explore those contradictions.
Imagine a cop says "stand over there" but a suspect doesn't know where they mean and doesn't move. Axon might well add "I pointed to the kerb" on the police report, which the police officer rubber stamps. Now the system has invented a clear non-complying suspect with nothing but their word to refute it.
Similarly shaky camera, limited field of view, poor audio, periods when cameras are off all add opportunities for camera footage to fail to prove what happened. Police reports have a lot of weight in court, they are treated as true by default.
We already know that police officers are not more reliable than the general public as eye witnesses and that eye witness reports are generally very unreliable as they are very susceptible to prompting bias. This seems like leaning in to prompt bias. The AI is now prompting the human rather than the other way around. This is perverse.
EU AI Act specifically calls out and forbids such applications. Of course, the state will do what the state will do, but there's an actual obstacle enshrined in law.
I worry that we’ve already created the AI apocalypse and that this is what it looks like, along with extremist magnification on social media.
I trust AI to be what is is- essentially a lot of math that classisfies and predicts stuff, usually words, that the prediction can be used generatively, and the classification stuff can be used to identify stuff in various media.
What I don’t trust is that people will use it responsibly. Hell, I don’t, when I’m vibe coding, but that’s on me.
People are venal and self absorbed and busy and lazy and all the other traits that lead to not using AI responsibly. And businesses are amoral (not immoral) and want the shortest path to revenue, with the least friction.
So of course police officers who want to be on patrol and did not sign up to spend countless hours on reports, are going to let the AI write it and call it good without proofreading.
We could pass a lot of laws trying to specify products that force police to act reliably, or we could maybe just pass a law that says that AI cannot be used to write police reports, but that clearly labeled AI generated transcriptions and summaries may be attached unedited to police reports, if and only if the original recordings are also preserved as evidence.
And police departments that keep body camera and car camera footage might be ease up on the report writing and only require officers to annotate it with their impressions, but otherwise let the record speak for itself.
As others point out, the actual bodycam footage will be definitively probative for the events it records. But there are plenty of cases where the report itself leads to later actions that may be tortious or criminal, and finding out who's to blame for the exact wording used is highly relevant.
Example: AI incorrectly reports that during A's arrest, A made incriminating allegations about B. Based on the report, the police get a warrant and search B's house. When it turns out B is innocent, B sues the department, and when the report turns up during discovery, we're off to the circus.