> According to the court documents, the Fargo detective working the case then looked at Lipps' social media accounts and Tennessee driver's license photo. In his charging document, the detective wrote that Lipps appeared to be the suspect based on facial features, body type and hairstyle and color.
> Once they were in hand, Fargo police met with him and Lipps at the Cass County jail on Dec. 19. She had already been in jail for more than five months. It was the first time police interviewed her.
How is this the fault of AI? It flagged a possible match. A live human detective confirmed it. And the criminal justice system, for reasons that have nothing to do with AI, let this woman sit in jail for 5 months before doing even interviewing her or doing any due diligence.
There's a reason why we don't let AI autonomously jail people. Instead of scapegoating an AI bogeyman, maybe we should look instead at the professional human-in-the-loop who shirked all responsibility, and a criminal justice system that thinks it is okay to jail people for 5 months before even starting to assess their guilt.
> How is this the fault of AI? It flagged a possible match. A live human detective confirmed it.
Because we're seeing the first instances of what reality looks like with AI in the hands of the average bear. Just like the excuse was "but the computer said it was correct," now we're just shifting to "but the AI said it was correct."
Don't underestimate how much authority and thinking people will delegate to machines. Not to mention the lengths they'll go to weasel out of taking responsibility for a screw up like this (saw another comment in this thread about the Chief of Police stepping down but it being framed as "retirement").
> I feel like I'm going crazy with this narrative.
We're only getting warmed up. There are programmers on HN that will take the output of their favorite AI, paste it and run it. And we're supposed to be the ones that know better.
What do you think an ordinary person is going to do in the presence of something that they can not relate to anything else except for an oracle, assuming they know the term? You put anything in there and out pops this extremely polished looking document, something that looks better than whatever you would put together yourself with a bunch of information on it that contains all kinds of juicy language geared up to make you believe the payload.
They're going to fall for it, without a second thought.
And they're going to draw consequences from it that you thought could use a little skepticism. Too late now.
You are exactly correct. Cops cannot be trusted. We spent a lot of time pointing that out in 2020. AI is the least of our problems with policing.
Unfortunately, a lot of people are certain it won't happen to them, and it has been practically impossible to establish any kind of accountability. It has only gotten worse since 2020.
As soon as we start to see a pattern of shitty vibe-coded software actually harming people via defects etc. (see: therac-25), I would hope that the conversation is about structural change to mitigate risk in aggregate rather than just punitive consequences for the individual programmers who are "responsible". The latter would be a fantastically stupid response and would do little or nothing to reduce future harm.
all accountability need not be punitive, we can certainly talk about systemic guardrails. What I find disbelief in, is someone saying the Chief of Police saying "We are not going to talk about that today?" is not the biggest scandal, but the AI is.
> someone saying the Chief of Police saying "We are not going to talk about that today?" is not the biggest scandal, but the AI is.
Who is this "someone"? OP's article and the discussion here are absolutely not neglecting the human factors and general
institutional failure that made this possible. But it's also true that without these "AI" tools, it would never have happened.
I mean, this is the USA we're talking about. Cops are given huge authority over everyone else, with poor accountability. AI just lets them pretend to be even less accountable. And by "pretend" I of course mean "get away with it".
You’re going crazy because up until this exact moment you’ve never had to confront the reality that these tools, placed into the hands of the common man, are viewed as authoritative and lack any accountability or consequence for misuse.
For anyone who has been victimized by law enforcement or governments before, we’ve been warning about this shit for decades. About the lack of consequence for police brutality. The lack of consequence for LPR abuse. The lack of consequence for facial recognition failures and AI mismatches.
You need to understand that by using these systems correctly and holding yourself accountable, you are in the minority. Most people do not think that critically, and are all too happy to finger the computer when things go badly.
And until you accept that, and work to actually hold folks accountable instead of deflecting blame away from the tool, then this won’t actually change.
It's called qualified immunity. Many support its repeal. I hope you join them, and convey the same to your local representatives and candidates. Until it is reformed few if any officers or administrators of criminal justice in the United States will ever feel any type of accountability.
Short of video evidence of blatant gun to the back of the head style homicide qualified immunity means most law enforcement officials are never held accountable for their miscarriages of justice. Criminal charges against officers are exceedingly rare. She should be able to sue this detective directly. Of course she can sue the government too, and should. But without any personal consequences for the people carrying out these acts, taxpayers will continue to bail out these practices without ever noticing. Your own government should not be a shield for a police officer who has violated you or your neighbors.
> Short of video evidence of blatant gun to the back of the head style homicide qualified immunity means most law enforcement officials are never held accountable for their miscarriages of justice.
This particular "AI bogeyman" isn't just AI; it's cops with AI and in particular cops with facial recognition tools, dragnet LPR surveillance tools, and all this other new technology that essentially picks somebody's name out of a hat to have their life temporarily (or [semi-]permanently) ruined by shithead cops who won't ever face any real accountability.
This keeps happening, and the reason it keeps happening is that shithead cops have these tools and are using them. Until we can find a reliable way to prevent this from happening, which may or may not be possible, cops who may or may not be shitheads should not have access to these tools.
They don't seem to give a single iota of a fuck about that when a private regular person has their money stolen or their car totaled by hit and run driver. Finding some innocent person to arrest would indicate they are at least pretending to give a fuck, yet they seem to only be bothered to even keep up appearances when it is the bank being robbed.
Let me help you out with this comprehension issue. The point of my comment is that I disagree with the apparent premise of the comment I replied to, which is that "AI" is some generic investigative tool that we can neatly snip out of the picture to blame this incident on human factors at the individual level ("the professional human-in-the-loop who shirked all responsibility"). Said comment also implies that people are fixating on the AI aspect of this issue while ignoring the human factors, which IMO is a strawman. To me, the existence of AI in its current incarnations and the ways in which law enforcement will inevitably abuse it are, together, inseparably, the problem. AI (in the most general sense) opens up entire new dimensions for potential abuse.
As a concrete example:
> And the criminal justice system, for reasons that have nothing to do with AI, let this woman sit in jail for 5 months before doing even interviewing her or doing any due diligence.
Let me state what should be obvious: without AI (as in, the facial recognition systems involved in this case), this woman would not have sat in jail for 5 months, or indeed for any length of time at all. So saying that it has "nothing to do with AI" is totally ridiculous.
It's not. This is just an acceleration in the unraveling of society facilitated by AI. As someone whose childhood included so many "robots will kill humans" books and movies, I am flabbergasted that the AI apocalypse will be dumb humans overtrusting faulty AI in important matters until everything falls apart.
Most humans cannot distinguish AI from actual intelligence. When you combine that with bureaucrats innate tendency to say, "Computer said so," you end up with bizarre situations like this. If a person had made this facial match, another human would have relentlessly jeered him. Since a computer running AI did it, no one even cared to think about it.
Computers are wildly dangerous, not because of anything innate but because of how humans act around them.
It's the fault of the tool because our society treats the tools as superior judgements than humans and to be trusted completely as a means of deflecting accountability - something any and every minority group has been warning about for fucking decades.
The reason everyone rushes to defend the tool's use is because holding humans accountable would mean throwing these tools out entirely in most cases, due to internal human biases and a decline in basic critical and cognitive thinking skills. The marketing has been the same since the 80s: the tool is superior (until it isn't), the tool shall be trusted completely (until it fails), the tool cannot make mistakes (until it does).
If folks actually listened to the victims of this shit, companies like Flock and Palantir would be gutted and their founders barred from any sort of office of responsibility, at minimum. The fact so many deflect blame from the tool like the marketing manual demands shows they don't actually give a shit about the humans wrapped up in the harms, or the misuse and misappropriation of these tools by persons wholly unaccountable under the law, but only about defending a shiny thing they personally like.
It could be the fault of the company that's selling this service. They often make wildly inaccurate claims about the utility and accuracy of their systems. [0]
> There's a reason why we don't let AI autonomously jail people.
Yes we do. [1]
> and a criminal justice system that thinks it is okay to jail people for 5 months before even starting to assess their guilt.
Her guilt was assessed. That's why she had no bail. It assessed it incorrectly, but the error is more complicated than your reaction implies.
There's no way this isn't a slam dunk case to sue the piss out of the Fargo Police, probably the US Marshals and maybe other orgs. The woman in the surveillance phone clearly looks way younger, among the many other obvious signs this woman didn't do it. I hope she wrings at least several million dollars out of the government.
You can be arrested, indicted, and held in jail on pretrial, and there is literally no recourse. There are many other ways jail can happen without due process. Where I live:
* Civil contempt. Absolutely immunity. No due process. Record is about 16 years. Having a bad day? Judge can toss you in jail.
* "Dangerous." Half a year. No due process. He-said she-said.
* "Insane." Psychiatric hold. Three days. Due process on paper, not in practice. Police in my town can and do use this if they don't like you.
Absolutely no recourse. You come out with a gap in income, employment, and, if you missed rent/mortgage, no home. Landlords will simply throw your stuff away too.
You're also basically damned if things do move forward, since from jail, you have no access to evidence, to internet (for legal research), and no reasonable way to recruit a lawyer (and, for most people, pay for one).
Can happen to anyone. Less common if you're rich and can afford a good lawyer, but far from uncommon.
>* "Insane." Psychiatric hold. Three days. Due process on paper, not in practice. Police in my town can and do use this if they don't like you.
A friend of mine was committed longer than 3 days without council or the ability to represent themselves in the hearing. Apparently the whole process of being committed is ex parte in practice in some states.
This is a bit hyperbolic and the exaggerations really undermine what I think is your broader point (that there is rarely recourse when you're held for short to moderate amounts of time). It is hard for me to believe that someone was held for 16 years on civil contempt without due process or that someone was held for half a year without due process after being deemed dangerous. The reason that is hard for me to believe is that the due process is implicit in the action you describe. Civil contempt is from a judge which implies that you're already in court - that's due process. Someone being labeled "dangerous" implies that a finding was made by a neutral party - that's due process.
Just because you disagree with the outcome doesn't mean that due process wasn't given.
Yeah it's "due process." In civil contempt the judge is a witness and prosecutor in the very "process" they're judging. That's the most perverted form of due process imaginable.
A judge should have to recuse themselves if they are acting as witness to the supposed infraction.
Civil contempt isn't some roving criminal charge that jumps out of the jury box randomly. It's meant to make somebody comply with a court order. Anybody in civil contempt holds the keys to the jailhouse door in their own hands, all they have to do is comply.
This statement should make you uncomfortable. It makes me uncomfortable because it is a pure expression of the power of the state. But it's still due process.
Qualified immunity doesn't apply to criminal cases. It is used to defend against civil suits. It is unfortunately very easy to find many cases where it leads to injustice. For example:
>...Abby Tiscareno, a licensed daycare provider in Utah, was wrongfully convicted of felony child abuse when a child under her care suffered brain hemorrhaging. After calling emergency services, subsequent medical tests supported these findings. However, during her trial, requested medical records from the Utah Division of Child and Family Services (DCFS) were not provided. It wasn’t until a civil suit that Ms. Tiscareno saw pathology reports suggesting the injury could have occurred outside of her care. She was granted a new trial and acquitted. Her subsequent lawsuit for due process violations, alleging that DCFS failed to provide exculpatory evidence, was dismissed due to lack of precedent indicating DCFS’s obligation to produce such evidence.
Off of taxpayer money sadly. Imo we really need a fix for this. When cops are grossly negligent the money should come out of their aggregate pension fund (or at least partially).
> we really need a fix for this. When cops are grossly negligent the money should come out of their aggregate pension fund
This is on us as voters. If we didn’t piss our pants every time a police union sneezed, we’d realize wholesale restarting police departments is precedents in even our largest cities.
Yes, this is the key point. Tax payers get a nice big bill while the people who caused the problem get a nice paid vacation while they conduct an internal "investigation" that typically finds they did nothing wrong.
Yeah, of course they need to held accountable, and we need to vote in people who will do so. What I'm suggesting is an alignment of incentives that will ensure that police will try to do their best to not be negligent.
Of course there's a balance that has to be struck so that police are empowered enough to act. So perhaps something like settlements against the police being 30% borne by the police pension fund and 70% by taxpayers is sufficient. I think this will also make police very enthusiastic about bodycams and holding each other accountable.
despite this being something practically everybody wants, the fact that it hasn't happened is not a coincidence and speaks to the power of police unions/guilds and their lobbying arms. outside a few toothless instances, those groups are extremely good at reframing these attempts and mobilizing their bases to vote against the broader public interest.
> despite this being something practically everybody wants,
No, everybody does not want police accountability. Half the population will fall on a grenade to prevent that. They know that the purpose of the police is to keep the undesirables in line, and they never envision that they will ever fall in that category.
oh, i generally don't disagree with you on that point; i specifically meant that when presented with the question "do you want your tax dollars to pay for police liabilities?" the answer is probably almost always "no".
Almost all taxpayer funded pension funds are already underfunded. It makes no difference if the funding decreases or increases, the government employee will still get their benefit. The government would have to go through bankruptcy to get the benefit amount reduced.
It literally doesn't matter -- you're focused on the wrong thing. She could be that woman's exact twin and it wouldn't matter. Spending six months in jail and losing your house, your car, and your dog with the flimsiest of evidence is ridiculous.
A lawsuit is exactly what matters. They learn only the hard way, and no other way. If you want them to not be ridiculous, a lawsuit with large punitive damages is the only practical way to get there.
I disagree. The city or state gets sued and they pay the result from the taxpayer funds and literally nobody learns anything, especially not the hard way. Everyone is so completely divorced, and in some cases immune, from consequences that this will change nothing.
imho the US Marshals are the only innocent party here, as my understanding is they don't do investigations and just serve warrants without any knowledge of the underlying case.
It is an AI error, but also an error on the part of the cops, the prosecutors, the judge, and the county sheriff (who is responsible for the jail inmates). I hope everyone involved in this travesty is sued into oblivion and unable to hide behind their immunity defenses. Facial recognition should never be the sole basis for a warrant.
> It is an AI error, but also an error on the part of the cops, the prosecutors, the judge, and the county sheriff
Yes, it's critical to remember that multiple parties can be at fault. In a case like this, it is true that
a) law enforcement misused a tool and demonstrated extreme negligence
b) the judiciary didn't catch this, which suggests systemic negligence there too when it comes to their oversight responsibilities
c) the company selling/providing this AI tool should have known it was likely to be misused and is responsible for damages caused by such predictable usage
We cannot have a just world until our laws and norms result in loss of jobs and legitimacy as punishment for this sort of normalized failure, from all three parties. Immunity is a failed experiment.
Even if she was a read ringer (clearly not the same person to any human who glances at the image), common sense should tell you that among 340,000,000 Americans there are a lot of lookalikes. Clearly there's a kind of stupid belief in the mystic powers of an AI and a callous disregard for the well being of suspects. No one should be dragged 1000 miles and held for months based on a facial match, especially when exculpatory evidence was easily available.
The software identified the person as Angela Lipps. According to the court documents, the Fargo detective working the case then looked at Lipps' social media accounts and Tennessee driver's license photo.
In his charging document, the detective wrote that Lipps appeared to be the suspect based on facial features, body type and hairstyle and color.
The software worked exactly as intended. It's a filtering tool that sifts through data for common patterns to provide leads, not matches. It raises a flag on persons of interest. You can be a "match" anywhere between 0 and 100% and only relative to some specific input (like that picture taken from the top of the woman at the teller). In that sens mismatches are within acceptable parameters and have been known to happen.
A "match" is a pronouncement ultimately made by the humans that uses the tool, after they've checked out the leads. Someone slept at the wheel here.
I followed the inquiry when it was ongoing — all of the depositions were live on YouTube. The level of both hubris and incompetence involved in that case was breathtaking.
Out of curiosity, was the guy known for being fast and loose with the rules? Put more simply, was he a good cop? Or did he have a history of being a rogue.
This problem predates modern AI. https://en.wikipedia.org/wiki/Computer_says_no is built upon the deliberate abdication of responsibility to processes that cannot be held accountable. AI is just letting them do it at scale.
That doesn't mean we should accept it from AI. We should fight the blind yielding to the facade of authority regardless of whether the decision was made by an AI or an insect landing on a teleprinter at the wrong time.
John Bryant, aka The Civil Rights Lawyer, recently did a piece about a similar case of mistaken identity. The consequences weren't as severe, but the willingness to trust the AI over any other evidence was the same:
In the video, it shows a police officer blindly trusting a casino's AI software, even when a cursory investigation should have given any reasonable person enough of a reason to question whether the man he arrested was the same man accused of a crime. (And then even after it was confirmed he was not, the prosecutor continued to charge him for trespassing!)
I really, really need folks to understand that deflecting blame away from the tool and trying to hold the human accountable feeds right into the marketing playbook of these companies in the first place.
The cops cannot be held accountable because the laws basically give them immunity. The politicians cannot be held accountable beyond being tossed out at the next election, because the laws otherwise give them immunity. The people operating the system cannot be held accountable, because the systems are marketed as authoritative despite being black boxes and lacking in transparency; they trusted the system just as they were told to, and thus cannot be held accountable.
And so when every human in the chain cannot be held accountable for these things, and the law prevents victims from receiving apologies, let alone recourse, then the tool and its maker is the only thing we can hold accountable. By deflecting blame away from the tools ("it wasn't AI, it was facial recognition"; "the human had to sign off on it"; "humans made the arrest, not machines"), you're protecting quite literally the only possible entity that could still potentially be held accountable: the dipshits making these stupid things and marketing them as superior and authoritative when compared to humans.
You want accountability? Start holding capital to account, and this shit falls away real fucking fast. Don't get lost in technical nuance over very real human issues.
>Unable to pay her bills from jail, she lost her home, her car and even her dog.
Fargo police say the bank fraud case is still under investigation and no arrests have been made.
Except in "Brazil" it was a mechanical error in a deterministic machine caused by an invasive outside actor. It would be reasonable to trust that the autotypewriter/printer would faithfully output the correct text.
Modern AI seems incapable of any respectable amount of accuracy or precision. Trusting that to destroy somebody's life is even more farcical than the oppressive police in "Brazil".
It’s obvious from the one photo they posted of the actual suspect that the lady they arrested is about 20-30 years older than the woman in the bank photo. The woman in the photo is maybe 25-30 years old, this grandma looks like she’s 65-70 (actual age of 50).
Absolutely ridiculous, I hope she wins her civil case.
Wait - what was the AI tool and how did it have her face to begin with? If small-town police are doing face-matching searches across national databases then nobody is safe because the number of false positives is going to be MASSIVE by sheer number of people being searched every day.
Pretend the tool is 99.999999% specific. If it searches every face in the USA you're still getting about 3 false positives PER SEARCH.
You will never have a criminal AI tool safe enough to apply at a national scale.
People will defend this, too, saying “well, she was eventually exonerated, right? So the system works!” Ignoring how she’ll never be fully reimbursed for the time, money, and grief of going through the system.
We also need to question how many people might go through the same process without eventual exoneration and how much going through this process costs individuals. Being falsely prosecuted comes usually imparts a permanent black mark in search results about the person (outside of places with sane laws like the EU) as well as causing stress or permanent injury.
Wrongly arrested individuals with mental disabilities have a history of physical abuse in jail potentially to the point of death.
Completely infuriating, but more of a commentary on the sad state of incompetent power-hungry law enforcement with tools they don't know how to use than the tools themselves.
Though, the question remains: are the tools built in such a way as to deceive the user into a false sense of trust or certainty?
But it’s glaringly obvious that if you build tools like this and give them to the US police this is the outcome you will get. The toolmakers deserve blame too.
> they don't know how to use than the tools themselves.
No, the tools work perfectly as they were design to work. The problem is that the tools are flawed.
Ultimately, every single of these decisions should be approved by a human, which should be responsible for the fuck up no matter what the consequences are.
> _Some_ of the blame lies on the UX here. It must.
No, the blame lies with the person or the group who approve the usage of these tools, without understanding their shortcomings.
>> are the tools built in such a way as to deceive the user into a false sense of trust or certainty? _Some_ of the blame lies on the UX here. It must.
> No, the blame lies with the person or the group who approve the usage of these tools, without understanding their shortcomings.
The person who approved the tools might've understood, but that doesn't mean the user understands. _Some_ of the reason why the user doesn't understand the shortcomings of the tool might be because of misleading UX.
I don’t know what tool they used, but it was very likely not an LLM. They probably have some database of drivers’ licenses and they ran a similarity search against the surveillance footage. This poor lady happened to be the top match.
Even if it also output a score, that score depends on how the model was trained. And the cops might ignore it anyways.
> are the tools built in such a way as to deceive the user into a false sense of trust or certainty? _Some_ of the blame lies on the UX here. It must.
Are AI code assist tools built in such a way as to deceive the user into a false sense of trust or certainty? Very much so (even if that isn't a primary objective).
Does any part of the blame lie on the UX if a dev submits a bad change? No, none.
You are ultimately, solely responsible for your work output, regardless of which tool you choose to use. If using your tool wrong means you make someone homeless, car-less, and also you kill their dog, then you should be a lot more cautious and perform a lot more verification than the average senior engineer.
I agree with all that. Maybe the word isn't "blame," then. Surely there must be some code, perhaps moral or ethical, but ideally more rigorously enforcible, which ought to prevent the development of intentionally deceiving tools. Sure you could say this about all software, but that which can cause actual physical harm ought to be held to a higher standard.
Yes, unfortunately technology is advancing faster than the average human brain evolves more neurons, so it will only become less comprehensible to the average person.
That's setting aside the tendency for police to hire from the left side of the bell curve to avoid independent thinkers that might question authority, refuse to do bad shit, etc.
I read the article and I don’t really understand… she was held in a jail in Tennessee but the article states they flew her to North Dakota? And somehow she’s a fugitive so that’s why she doesn’t get bail? but she’s a fugitive held in her own state in a holding facility? But then when they release her, she’s in North Dakota? So if some state says you’re a fugitive your home state will just hold you in jail until they come and put you on an airplane? Is that correct?
I think you have the interpretation correct. It seems like any state can say you're a fugitive from their state and now you have even fewer rights. Every day I learn some new fact about "justice" in the United States.
It's annoying that both articles are calling this AI error. This was human error, the police did the wrong thing and the people of Fargo will end up paying for this fuckup.
I would argue it was both. No doubt this company was marketing it in a way to make it seem very reliable. And all of the procedural things afterwards made the error so much more damaging.
But imo this is why local police departments should not have access to this kind of tool. It is too powerful, and the statistical interpretation is too complicated for random North Dakota cops to use responsibly. Neither the company nor the PD have an incentive to be careful.
It's not an AI error. It's a human error in mis-using AI in this way. Saying it's an AI error is like saying a hole in your drywall is a hammer error.
Unfortunately we'll probably see a trend of people using AI and then blaming AI for cases where they mis-used AI in roles it's not good for or failed to review or monitor the AI.
It's both. It's good to acknowledge that AI is easy to misuse in this manner but it doesn't detract from the fact that the ultimate responsibility lies in those that should be verifying the tool output.
There is far too little skepticism around the magic box that solves all problems which is causing issues like this. It's not the fault of the AI (as if it could be assigned liability) for being misused, but this kind of misuse is far too common right now so scare stories like this are helpful and we should highlight the use of AI in mistakes like this.
I hate this headline (not blaming submitter). Police incompetence and negligence jailed her for months and left her stranded in a North Dakota winter. The AI is no more responsible than the cars and airplanes they used.
Edit: this is in reference to the original headline "AI error jails innocent grandmother for months in North Dakota fraud case" not the revised title that it was changed to.
If it didn't erase accountability, how would it create any value?
Many people are treating this as a matter of philosophy, which it isn't.
At a primitive, physiological level if you delegate to AI and most of the time you don't get in trouble for it, the resulting relationship you have with the AI could only be called "trust".
If you're expected to be 40% more productive at your job, your employer is making it crystal clear that you will trust the AI or you will be fired. Even if nobody ever said it, the sales pitch is that AI does the work and people are mostly there to be their servants whose role is to keep them fed with decisions we want made but don't want to be responsible for making.
Your picking apart the words doesn't matter if police are more incompetent with AI than without it. AI being the catalyst to a worse society is a more interesting and worthwhile topic than whether "AI is responsible" is the right way to phrase it.
If you make the AI software, then your software malfunctioned.
If the laser printer screws up a page in the middle of the document, and the user doesn't catch it and includes it in the board of directors binder, the laser printer still malfunctioned.
We are rapidly becoming a world where every person is one inscrutable LLM decision from having their life ruined with no recourse.
This type of incident isn't new and is only going to get worse. The problem is our governments are doing absolutely nothing about it. I'll give two examples:
1. Hertz implemented a system where they falsely reported cars as being stolen. People were arrested and went to jail for rental cars that were sitting in the Hertz lot. Hertz ultimately had to pay $168 million in a settlement [1]. That's insufficient. If I, as an ordinary citizen, make a false police report that somebody stole my car I can be criminally charged. And rightly so. People should go to jail for this and it will continue until they do. These fines and settlements are just the cost of doing business; and
2. The UK government contracted Fujitsu to produce a new system for their post offices. That system was allowed to produce criminal charges for fraud that were completely false. People committed suicide over this. This went on for what? A decade or more? But resuted in a parliamentary inquiry and settlements. It's known as the British Post Office scandal [2]. Again, people should go to jail for this.
The choice we as a society face is whether to have automation improve all of our lives by raising everyone's standard of living and allowing us to do less work and less menial work or do we allow automation to further suppress wages so the Epstein class can be slightly more wealthy.
Because it has an updating-feed-like structure, in which new items can appear.
Knowing that there are (N) new items is so useful (to some people), that as far back as the 1990s, we developed technology called "RSS" to give you this superpower over a website that doesn't provide anything of the sort. One that simply updates with new stuff when you hit refresh, with no UI to indicate what is new/changed.
> Once they were in hand, Fargo police met with him and Lipps at the Cass County jail on Dec. 19. She had already been in jail for more than five months. It was the first time police interviewed her.
How is this the fault of AI? It flagged a possible match. A live human detective confirmed it. And the criminal justice system, for reasons that have nothing to do with AI, let this woman sit in jail for 5 months before doing even interviewing her or doing any due diligence.
There's a reason why we don't let AI autonomously jail people. Instead of scapegoating an AI bogeyman, maybe we should look instead at the professional human-in-the-loop who shirked all responsibility, and a criminal justice system that thinks it is okay to jail people for 5 months before even starting to assess their guilt.
Because we're seeing the first instances of what reality looks like with AI in the hands of the average bear. Just like the excuse was "but the computer said it was correct," now we're just shifting to "but the AI said it was correct."
Don't underestimate how much authority and thinking people will delegate to machines. Not to mention the lengths they'll go to weasel out of taking responsibility for a screw up like this (saw another comment in this thread about the Chief of Police stepping down but it being framed as "retirement").
Why are cops not treated the same way? OP is right, AI is totally irrelevant in this story.
If the point is "cops can't be trusted". Why do they have GUNS?! AI is the least of your problems.
I feel like I'm going crazy with this narrative.
We're only getting warmed up. There are programmers on HN that will take the output of their favorite AI, paste it and run it. And we're supposed to be the ones that know better.
What do you think an ordinary person is going to do in the presence of something that they can not relate to anything else except for an oracle, assuming they know the term? You put anything in there and out pops this extremely polished looking document, something that looks better than whatever you would put together yourself with a bunch of information on it that contains all kinds of juicy language geared up to make you believe the payload.
They're going to fall for it, without a second thought.
And they're going to draw consequences from it that you thought could use a little skepticism. Too late now.
Unfortunately, a lot of people are certain it won't happen to them, and it has been practically impossible to establish any kind of accountability. It has only gotten worse since 2020.
Who is this "someone"? OP's article and the discussion here are absolutely not neglecting the human factors and general institutional failure that made this possible. But it's also true that without these "AI" tools, it would never have happened.
For anyone who has been victimized by law enforcement or governments before, we’ve been warning about this shit for decades. About the lack of consequence for police brutality. The lack of consequence for LPR abuse. The lack of consequence for facial recognition failures and AI mismatches.
You need to understand that by using these systems correctly and holding yourself accountable, you are in the minority. Most people do not think that critically, and are all too happy to finger the computer when things go badly.
And until you accept that, and work to actually hold folks accountable instead of deflecting blame away from the tool, then this won’t actually change.
Short of video evidence of blatant gun to the back of the head style homicide qualified immunity means most law enforcement officials are never held accountable for their miscarriages of justice. Criminal charges against officers are exceedingly rare. She should be able to sue this detective directly. Of course she can sue the government too, and should. But without any personal consequences for the people carrying out these acts, taxpayers will continue to bail out these practices without ever noticing. Your own government should not be a shield for a police officer who has violated you or your neighbors.
And frequently not even then.
This keeps happening, and the reason it keeps happening is that shithead cops have these tools and are using them. Until we can find a reliable way to prevent this from happening, which may or may not be possible, cops who may or may not be shitheads should not have access to these tools.
As a concrete example:
> And the criminal justice system, for reasons that have nothing to do with AI, let this woman sit in jail for 5 months before doing even interviewing her or doing any due diligence.
Let me state what should be obvious: without AI (as in, the facial recognition systems involved in this case), this woman would not have sat in jail for 5 months, or indeed for any length of time at all. So saying that it has "nothing to do with AI" is totally ridiculous.
>> How is this the fault of AI?
> This particular "AI bogeyman" isn't just AI; it's cops with AI
Most humans cannot distinguish AI from actual intelligence. When you combine that with bureaucrats innate tendency to say, "Computer said so," you end up with bizarre situations like this. If a person had made this facial match, another human would have relentlessly jeered him. Since a computer running AI did it, no one even cared to think about it.
Computers are wildly dangerous, not because of anything innate but because of how humans act around them.
The false positive rate combined with scanning millions of pictures might make the chance of arresting the wrong person really high.
The reason everyone rushes to defend the tool's use is because holding humans accountable would mean throwing these tools out entirely in most cases, due to internal human biases and a decline in basic critical and cognitive thinking skills. The marketing has been the same since the 80s: the tool is superior (until it isn't), the tool shall be trusted completely (until it fails), the tool cannot make mistakes (until it does).
If folks actually listened to the victims of this shit, companies like Flock and Palantir would be gutted and their founders barred from any sort of office of responsibility, at minimum. The fact so many deflect blame from the tool like the marketing manual demands shows they don't actually give a shit about the humans wrapped up in the harms, or the misuse and misappropriation of these tools by persons wholly unaccountable under the law, but only about defending a shiny thing they personally like.
It could be the fault of the company that's selling this service. They often make wildly inaccurate claims about the utility and accuracy of their systems. [0]
> There's a reason why we don't let AI autonomously jail people.
Yes we do. [1]
> and a criminal justice system that thinks it is okay to jail people for 5 months before even starting to assess their guilt.
Her guilt was assessed. That's why she had no bail. It assessed it incorrectly, but the error is more complicated than your reaction implies.
[0]: https://thisisreno.com/2026/03/lawsuit-reno-police-ai-polici...
[1]: https://projects.tampabay.com/projects/2020/investigations/p...
https://abovethelaw.com/2016/02/criminally-yours-indicting-a...
You can be arrested, indicted, and held in jail on pretrial, and there is literally no recourse. There are many other ways jail can happen without due process. Where I live:
* Civil contempt. Absolutely immunity. No due process. Record is about 16 years. Having a bad day? Judge can toss you in jail.
* "Dangerous." Half a year. No due process. He-said she-said.
* "Insane." Psychiatric hold. Three days. Due process on paper, not in practice. Police in my town can and do use this if they don't like you.
Absolutely no recourse. You come out with a gap in income, employment, and, if you missed rent/mortgage, no home. Landlords will simply throw your stuff away too.
You're also basically damned if things do move forward, since from jail, you have no access to evidence, to internet (for legal research), and no reasonable way to recruit a lawyer (and, for most people, pay for one).
Can happen to anyone. Less common if you're rich and can afford a good lawyer, but far from uncommon.
Qualified immunity protects individuals, not departments, from liability.
The particular thread (in this thread) that I was responding to:
>> I hope she wrings at least several million dollars out of the government.
> With all the lovely qualified immunity doctrine? That's wishful thinking.
I was responding to the claim that qualified immunity protected the government, it does not.
A friend of mine was committed longer than 3 days without council or the ability to represent themselves in the hearing. Apparently the whole process of being committed is ex parte in practice in some states.
Just because you disagree with the outcome doesn't mean that due process wasn't given.
A judge should have to recuse themselves if they are acting as witness to the supposed infraction.
This statement should make you uncomfortable. It makes me uncomfortable because it is a pure expression of the power of the state. But it's still due process.
>...Abby Tiscareno, a licensed daycare provider in Utah, was wrongfully convicted of felony child abuse when a child under her care suffered brain hemorrhaging. After calling emergency services, subsequent medical tests supported these findings. However, during her trial, requested medical records from the Utah Division of Child and Family Services (DCFS) were not provided. It wasn’t until a civil suit that Ms. Tiscareno saw pathology reports suggesting the injury could have occurred outside of her care. She was granted a new trial and acquitted. Her subsequent lawsuit for due process violations, alleging that DCFS failed to provide exculpatory evidence, was dismissed due to lack of precedent indicating DCFS’s obligation to produce such evidence.
https://innocenceproject.org/news/what-you-need-to-know-abou...
This is on us as voters. If we didn’t piss our pants every time a police union sneezed, we’d realize wholesale restarting police departments is precedents in even our largest cities.
As long as you keep electing clowns that let the police do whatever they want, the police will... Do whatever they want.
Of course there's a balance that has to be struck so that police are empowered enough to act. So perhaps something like settlements against the police being 30% borne by the police pension fund and 70% by taxpayers is sufficient. I think this will also make police very enthusiastic about bodycams and holding each other accountable.
it sucks.
No, everybody does not want police accountability. Half the population will fall on a grenade to prevent that. They know that the purpose of the police is to keep the undesirables in line, and they never envision that they will ever fall in that category.
The brutality is the point for them.
which the citizens end up footing the bill for. yay.
Who stole her dog?!
> Her bank records showed she was more than 1,200 miles away, at home in Tennessee at the same time police claimed she was in Fargo committing fraud.
> Unable to pay her bills from jail, she lost her home, her car and even her dog
Yes, it's critical to remember that multiple parties can be at fault. In a case like this, it is true that
a) law enforcement misused a tool and demonstrated extreme negligence
b) the judiciary didn't catch this, which suggests systemic negligence there too when it comes to their oversight responsibilities
c) the company selling/providing this AI tool should have known it was likely to be misused and is responsible for damages caused by such predictable usage
We cannot have a just world until our laws and norms result in loss of jobs and legitimacy as punishment for this sort of normalized failure, from all three parties. Immunity is a failed experiment.
The software identified the person as Angela Lipps. According to the court documents, the Fargo detective working the case then looked at Lipps' social media accounts and Tennessee driver's license photo.
In his charging document, the detective wrote that Lipps appeared to be the suspect based on facial features, body type and hairstyle and color.
The software worked exactly as intended. It's a filtering tool that sifts through data for common patterns to provide leads, not matches. It raises a flag on persons of interest. You can be a "match" anywhere between 0 and 100% and only relative to some specific input (like that picture taken from the top of the woman at the teller). In that sens mismatches are within acceptable parameters and have been known to happen.
A "match" is a pronouncement ultimately made by the humans that uses the tool, after they've checked out the leads. Someone slept at the wheel here.
Same comment?
Also me, reading further: Uh-oh.
The chief of police also resigned today; wouldn't be shocked if this was part of the reasoning.
Source?
That said, it's portrayed as a retirement, and doesn't seem to give any hints that it's connected.
That doesn't mean we should accept it from AI. We should fight the blind yielding to the facade of authority regardless of whether the decision was made by an AI or an insect landing on a teleprinter at the wrong time.
https://thecivilrightslawyer.com/2026/03/11/ai-software-tell...
In the video, it shows a police officer blindly trusting a casino's AI software, even when a cursory investigation should have given any reasonable person enough of a reason to question whether the man he arrested was the same man accused of a crime. (And then even after it was confirmed he was not, the prosecutor continued to charge him for trespassing!)
The cops cannot be held accountable because the laws basically give them immunity. The politicians cannot be held accountable beyond being tossed out at the next election, because the laws otherwise give them immunity. The people operating the system cannot be held accountable, because the systems are marketed as authoritative despite being black boxes and lacking in transparency; they trusted the system just as they were told to, and thus cannot be held accountable.
And so when every human in the chain cannot be held accountable for these things, and the law prevents victims from receiving apologies, let alone recourse, then the tool and its maker is the only thing we can hold accountable. By deflecting blame away from the tools ("it wasn't AI, it was facial recognition"; "the human had to sign off on it"; "humans made the arrest, not machines"), you're protecting quite literally the only possible entity that could still potentially be held accountable: the dipshits making these stupid things and marketing them as superior and authoritative when compared to humans.
You want accountability? Start holding capital to account, and this shit falls away real fucking fast. Don't get lost in technical nuance over very real human issues.
I smell a lawsuit
Modern AI seems incapable of any respectable amount of accuracy or precision. Trusting that to destroy somebody's life is even more farcical than the oppressive police in "Brazil".
https://nob.cs.ucdavis.edu/classes/ecs153-2019-04/readings/c...
“A computer can never be held accountable, therefore a computer must never make a management decision.”
(https://www.ibm.com/think/insights/ai-decision-making-where-...)
End qualified immunity and see how fast cops start to do their jobs with care.
Winning a lawsuit literally ends in your own community members (not the cops) paying the bill.
/i
Absolutely ridiculous, I hope she wins her civil case.
Pretend the tool is 99.999999% specific. If it searches every face in the USA you're still getting about 3 false positives PER SEARCH.
You will never have a criminal AI tool safe enough to apply at a national scale.
This is from the Sixth Amendment. Where the rubber hits the road is what “speedy” means.
Wrongly arrested individuals with mental disabilities have a history of physical abuse in jail potentially to the point of death.
Though, the question remains: are the tools built in such a way as to deceive the user into a false sense of trust or certainty?
_Some_ of the blame lies on the UX here. It must.
The cops need to be held accountable.
But it’s glaringly obvious that if you build tools like this and give them to the US police this is the outcome you will get. The toolmakers deserve blame too.
No, the tools work perfectly as they were design to work. The problem is that the tools are flawed.
Ultimately, every single of these decisions should be approved by a human, which should be responsible for the fuck up no matter what the consequences are.
> _Some_ of the blame lies on the UX here. It must.
No, the blame lies with the person or the group who approve the usage of these tools, without understanding their shortcomings.
> No, the blame lies with the person or the group who approve the usage of these tools, without understanding their shortcomings.
The person who approved the tools might've understood, but that doesn't mean the user understands. _Some_ of the reason why the user doesn't understand the shortcomings of the tool might be because of misleading UX.
New LLM-related AIs are all supremely confident in every assertion, no matter how wrong.
Even if it also output a score, that score depends on how the model was trained. And the cops might ignore it anyways.
Are AI code assist tools built in such a way as to deceive the user into a false sense of trust or certainty? Very much so (even if that isn't a primary objective).
Does any part of the blame lie on the UX if a dev submits a bad change? No, none.
You are ultimately, solely responsible for your work output, regardless of which tool you choose to use. If using your tool wrong means you make someone homeless, car-less, and also you kill their dog, then you should be a lot more cautious and perform a lot more verification than the average senior engineer.
That's setting aside the tendency for police to hire from the left side of the bell curve to avoid independent thinkers that might question authority, refuse to do bad shit, etc.
Sales will sell the dream, who cares if the real world outcomes don’t align?
https://www.theguardian.com/us-news/2026/mar/12/tennessee-gr... - Another article on this without a paywall.
It's annoying that both articles are calling this AI error. This was human error, the police did the wrong thing and the people of Fargo will end up paying for this fuckup.
But imo this is why local police departments should not have access to this kind of tool. It is too powerful, and the statistical interpretation is too complicated for random North Dakota cops to use responsibly. Neither the company nor the PD have an incentive to be careful.
When I load this URL I get "One more step Please complete the security check to access" and I cannot get past the archive.is computational paywall.
But the guardian article actually has text! Thanks.
Unfortunately we'll probably see a trend of people using AI and then blaming AI for cases where they mis-used AI in roles it's not good for or failed to review or monitor the AI.
There is far too little skepticism around the magic box that solves all problems which is causing issues like this. It's not the fault of the AI (as if it could be assigned liability) for being misused, but this kind of misuse is far too common right now so scare stories like this are helpful and we should highlight the use of AI in mistakes like this.
Edit: this is in reference to the original headline "AI error jails innocent grandmother for months in North Dakota fraud case" not the revised title that it was changed to.
The AI made the call and humans licked its butthole
Many people are treating this as a matter of philosophy, which it isn't.
At a primitive, physiological level if you delegate to AI and most of the time you don't get in trouble for it, the resulting relationship you have with the AI could only be called "trust".
If you're expected to be 40% more productive at your job, your employer is making it crystal clear that you will trust the AI or you will be fired. Even if nobody ever said it, the sales pitch is that AI does the work and people are mostly there to be their servants whose role is to keep them fed with decisions we want made but don't want to be responsible for making.
Yes, that's what the OPs "incompetence and negligence" referred to.
If the laser printer screws up a page in the middle of the document, and the user doesn't catch it and includes it in the board of directors binder, the laser printer still malfunctioned.
This type of incident isn't new and is only going to get worse. The problem is our governments are doing absolutely nothing about it. I'll give two examples:
1. Hertz implemented a system where they falsely reported cars as being stolen. People were arrested and went to jail for rental cars that were sitting in the Hertz lot. Hertz ultimately had to pay $168 million in a settlement [1]. That's insufficient. If I, as an ordinary citizen, make a false police report that somebody stole my car I can be criminally charged. And rightly so. People should go to jail for this and it will continue until they do. These fines and settlements are just the cost of doing business; and
2. The UK government contracted Fujitsu to produce a new system for their post offices. That system was allowed to produce criminal charges for fraud that were completely false. People committed suicide over this. This went on for what? A decade or more? But resuted in a parliamentary inquiry and settlements. It's known as the British Post Office scandal [2]. Again, people should go to jail for this.
The choice we as a society face is whether to have automation improve all of our lives by raising everyone's standard of living and allowing us to do less work and less menial work or do we allow automation to further suppress wages so the Epstein class can be slightly more wealthy.
[1]: https://www.npr.org/2022/12/06/1140998674/hertz-false-accusa...
[2]: https://en.wikipedia.org/wiki/British_Post_Office_scandal
Knowing that there are (N) new items is so useful (to some people), that as far back as the 1990s, we developed technology called "RSS" to give you this superpower over a website that doesn't provide anything of the sort. One that simply updates with new stuff when you hit refresh, with no UI to indicate what is new/changed.