Arrested by AI: When Police Ignored Standards After AI Facial-Recognition Matches (msn.com) 55
A county transit police detective fed a poor-quality image to an AI-powered facial recognition program, remembers the Washington Post, leading to the arrest of "Christopher Gatlin, a 29-year-old father of four who had no apparent ties to the crime scene nor a history of violent offenses." He was unable to post the $75,000 cash bond required, and "jailed for a crime he says he didn't commit, it would take Gatlin more than two years to clear his name."
A Washington Post investigation into police use of facial recognition software found that law enforcement agencies across the nation are using the artificial intelligence tools in a way they were never intended to be used: as a shortcut to finding and arresting suspects without other evidence... The Post reviewed documents from 23 police departments where detailed records about facial recognition use are available and found that 15 departments spanning 12 states arrested suspects identified through AI matches without any independent evidence connecting them to the crime — in most cases contradicting their own internal policies requiring officers to corroborate all leads found through AI. Some law enforcement officers using the technology appeared to abandon traditional policing standards and treat software suggestions as facts, The Post found. One police report referred to an uncorroborated AI result as a "100% match." Another said police used the software to "immediately and unquestionably" identify a suspected thief.
Gatlin is one of at least eight people wrongfully arrested in the United States after being identified through facial recognition... All of the cases were eventually dismissed. Police probably could have eliminated most of the people as suspects before their arrest through basic police work, such as checking alibis, comparing tattoos, or, in one case, following DNA and fingerprint evidence left at the scene.
Some statistics from the article about the eight wrongfully-arrested people:
Gatlin is one of at least eight people wrongfully arrested in the United States after being identified through facial recognition... All of the cases were eventually dismissed. Police probably could have eliminated most of the people as suspects before their arrest through basic police work, such as checking alibis, comparing tattoos, or, in one case, following DNA and fingerprint evidence left at the scene.
Some statistics from the article about the eight wrongfully-arrested people:
- In six cases police failed to check alibis
- In two cases police ignored evidence that contradicted their theory
- In five cases police failed to collect key pieces of evidence
- In three cases police ignored suspects' physical characteristics
- In six cases police relied on problematic witness statements
The article provides two examples of police departments forced to pay $300,000 settlements after wrongful arrests caused by AI mismatches. But "In interviews with The Post, all eight people known to have been wrongly arrested said the experience had left permanent scars: lost jobs, damaged relationships, missed payments on car and home loans. Some said they had to send their children to counseling to work through the trauma of watching their mother or father get arrested on the front lawn.
"Most said they also developed a fear of police."
Bad cops gonna bad cop (Score:5, Insightful)
AI can't fix stupid or lazy.
Police procedures exist. They weren't followed. Shit went bad.
But... but.. AI! yeah, no. Just bad cops.
Bad cops aren't the problem (Score:5, Insightful)
But this isn't even a few bad apples. The barrel the apples are put in is full of mold and filth.
For starters Go on YouTube and watch some police training videos. They are virtually indistinguishable from military training videos. We train cops like they're in a war zone and then we get surprised when they're violent. We tell them over and over again that everyone is out to get them and we're surprised their trigger happy. That's a systemic problem.
In this case cops want to do what everybody does and make their numbers look good and what better way to do that then with fishing expeditions that let them randomly arrest minor offenders. Meanwhile the taxpayer spends hundreds of thousands if not millions locking up those minor offenders and our system grinds them into pulp turning them into worse criminals
Re: (Score:3)
It's not an individual thing or a few bad apples
Yes, but those apples are bad with or without AI. People are arrested for BS reasons all the time.
The whole system is rotten to the core.
Of course, but that isn't caused by AI. It's caused by a "lock'em up" mentality with little concern if the person being locked up is guilty. We just had a national election, and criminal justice reform wasn't even on the agenda, so don't expect anything to change.
In theory, AI should help. Fewer innocent people will end up in prison if the guilty can be more easily identified.
The proper way to use facial recogn
You're missing the point (Score:2, Insightful)
You can't fix this by taking out a few bad apples because the system creates bad apples by design. It's like trying to grow food in soil poisoned by nicotine. It's going to be in all the food and the food
Re: (Score:2)
The system turns cops into bad apples by definition.
That's a silly exaggeration. There is no such "definition".
It's like trying to grow food in soil poisoned by nicotine. It's going to be in all the food and the food is going to make you sick
Nicotine rapidly breaks down in soil and is not absorbed by roots. Adding nicotine to soil won't make you sick.
Re:You're missing the point (Score:4, Informative)
The system turns cops into bad apples by definition.
That's a silly exaggeration. There is no such "definition".
Hardly.
Ever wonder why we don't hear the "good cop" calling out the "bad cops"? Because their union frowns upon such things, and if the 'good cops' do call out other cops, they get shunned for it, basically for life.
So rsilvergun is quite correct that the system turns cops into bad apples, "by definition", regardless if you consider it an "exaggeration".
Now, no doubt you'll get all butthurt about someone not agreeing 100% with you, so I await your reply claiming everyone else but you is wrong using nonsensical arguments in the most convoluted way possible....
Re: (Score:2)
The system turns cops into bad apples by definition.
That's a silly exaggeration. There is no such "definition".
Hardly.
Ever wonder why we don't hear the "good cop" calling out the "bad cops"? Because their union frowns upon such things, and if the 'good cops' do call out other cops, they get shunned for it, basically for life.
Ever wonder why “good cops” don’t get the recognition they deserve? Because they’re too busy covering the ass of bad cops.
If there is no actual written definition, rule, or policy that encourages or even mandates reporting of bad/illegal behavior, then that’s not a “union”.
It’s a fucking gang. Call it what it is. Then DO something about it.
Re: (Score:1)
A lesson everyone in America needs to learn eventually.
Re: (Score:2)
The definition is in the full expression: A few bad apples spoil the bunch.
We know there are bad apples in the police. We've been talking about them for decades to the point we're just shrugging them off. But their presence corrupts the rest of the department by association - the same way a moldy apple will infect the apples around it until they're all moldy. The healthy apples don't cure the sick one.
Predisposition to arrest (Score:2)
Scenarios after the police get a face detection match
Police arrest the person cases
Scenario 1) Police arrest the person, the prosecutor finds out the person is not the suspect and gets them released before trial
Scenario 2) Police arrest the person, the person is prosecuted and goes to trial and is found guilty or not guilty
Police do not arrest the person
Scenario 3) Police do not arrest the person, the person is never arrested again and not ever tired on this or any other possible crimes
Scenario 4) Police do
Re:Bad cops gonna bad cop (Score:5, Insightful)
I'm curious if these investigators kept their jobs, and if not, what city they moved to to get re-hired.
Re: (Score:2)
Of course they kept their jobs. Dirty cops basically have to brazenly murder someone on-camera in order to get punished in any way whatsoever. And even when openly misbehaving and abusing the public on-camera, they get off scot-free or with little more than a slap on the wrist the vast majority of the time.
Dirty cops like these who "merely" try to send someone up the river who is not, in fact, guilty? Have that particular sort of shitheel EVER been appropriately* punished?
(* IE: Terminated, with cause s
Re: (Score:2)
"AI can't fix stupid or lazy police."
FTFY
Re:Bad cops gonna bad cop (Score:4, Insightful)
If you want to blame someone, you should blame the supplier of these "superhuman" AIs. Maybe put a few of them in jail, teach them to dogfood their own software.
Re:Bad cops gonna bad cop (Score:4, Insightful)
AI can, however, be a tremendous APPEAL TO AUTHORITY.
The almighty computer, that never gets anything wrong. Just shrug and point to the piece of paper in your hand. Afterall, the AI printed it.
They can hide their lazy police work behind the bad AI they are using.
Re: (Score:2)
AI can't fix stupid or lazy.
No but courts can. Seems they don' work either. So the whole system is fucked up.
Re: (Score:2)
Re: (Score:2)
You're not supposed to rely on "the computer said so" has done virtually nothing to prevent the flood of "the computer said so" that has been pouring out ever since... computers.
At this point anyone pretending it's a solution is no better than the clowns spouting it.
Never intended? LOL (Score:3)
"...law enforcement agencies across the nation are using the artificial intelligence tools in a way they were never intended to be used: as a shortcut to finding and arresting suspects without other evidence..."
That's EXACTLY how they are intended to be used. What's next, will they find that mace and tasers are not intended to impose pain and suffering by police delivering street justice? Do they think we are stupid?
The people who develop and fund these systems know exactly how they will be used, they just deny it because the intention is to violate people's rights.
Re: (Score:1)
I think its also the same problem when we have witnesses try to point out a person from a lineup, when in reality this is very flawed because people are as bad as the facial recognition algorithms, which is why we have stunt doubles and celebrity look-alikes.
Re: (Score:2, Interesting)
because people are as bad as the facial recognition algorithms
People are much worse than facial recognition algorithms.
Face recognition like drug sniffing dogs (Score:2)
It amazes me just how much we give up. We give up tons and tons of our rights because we're all terrified of drugs. California just passed a law throwing minor shoplifters into prison for 3 years. So somebody steals a hundred bucks worth of crap from Walmart and now the taxpayers on the hook for approximately $400,000 in costs.
Re: (Score:3)
More and more people are waking up to the reality of the 'war and drugs', 'tough on crime' nonsense, and other right-wing bullshit. We know from decades of polling that progressive policies are extremely popular, even among so-called 'conservatives', provided you don't call them progressive. The outrage on the right is justified, it's just misplaced. History tell us that they won't stay deluded forever and, if promises are kept, they won't stay deluded for much longer.
At the moment, too many of them are
Re: (Score:2)
What law throws people who steal $100 worth of crap from Walmart in prison for 3 years?
Assuming it is just ordinary shoplifting I mean, without violence or other aggravating circumstances.
Assuming such law exists, what are the precedents? It is common for laws to mention a heavy penalty for seemingly small crimes, but the actual judgment is typically much more reasonable unless there is something much bigger behind it.
It doesn't mean there isn't a problem with the police, in fact, there is a problem with th
Re: Face recognition like drug sniffing dogs (Score:2)
"Most said they also developed a fear of police." (Score:3, Interesting)
So it's working as intended, then?
Linking to MSN for WaPo? (Score:2)
Why does this entry refer to a Washington Post article but links to MSN?
Here is the actual WaPo link:
https://www.washingtonpost.com... [washingtonpost.com]
Re: (Score:3)
Because WaPo is paywalled.
PIBKAC (Score:2)
The problem is not the facial recognition software. The problem is between the keyboard and the chair.
I think the police should use facial recognition software, if needed, as STEP 1 OF MANY!
Then, compare photos and physical characteristics. Then do something called detective work.
Re: (Score:3)
I'd be wary of it being step 1, but yes, there should have been significantly more steps in the detective work.
A problem is that they started with a hypothesis and then tried to prove that hypothesis correct. Confirmation bias is guaranteed in such an approach, however much care is taken.
Police should systematically gather all evidence they can, of all types, but then record what the evidence precludes, not just what the evidence allows. This is key in sanity-checking hypotheses - a hypothesis your evidence
Corrupt and/or incompetent, probably both (Score:3)
the software spat out the names and photos of several people deemed to resemble one of the attackers, whose face was hooded by a winter coat and partially obscured by a surgical mask.
This is all you need to know about America, its cops and its "justice" system.
The improper photo lineup that led to Chris Gatlin's arrest. Feldman's statement was later thrown out by a judge, who said the officers failed to conduct a fair and impartial photo lineup prior to arresting Christopher Gatlin.
It's safe to say that the incompetent cops went even further and rigged the photo lineup to attempt to confirm their ridiculous AI fishing expedition. Bad on top of bad on top of really bad.
Gatlin’s photo was shown to the victim of the assault, along with photos of five other individuals – even though the victim had said that, because of the injuries he sustained, he couldn’t remember what his attackers looked like. Eventually Gatlin’s photo was selected. As the Post reports – and as the lead detective on the case later admitted in court – the photo lineup was handled improperly. The detective influenced the victim’s selection.
Police in Woodbridge, New Jersey, were so intent on building a case against Nijeer Parks, a robbery suspect, that they cast aside DNA and fingerprint evidence pointing to another potential suspect.
Somehow they found even worse cops on another case. They willfully ignored actual evidence.
We tried and failed years ago (Score:5, Insightful)
More than 10 years ago, my company tried to do facial recognition in an airport in Europe, for their security service. Alas, they didn't know about the "birthday paradox", and tried to match about 1,000 criminals against several thousand passengers. They shut it down when the system identified someone's grandmother as a male member of the Baader-Meinhof gang.
The (birthday) paradox is caused by trying to match each passenger against 1,000 criminals, not just one. Even with only a 1% error rate, there will be 10 false positives and negatives per passenger. And we don't have anything like a 1% error rate.
We need a 1/infinity error rate (:-)) Otherwise innocent grandmothers will be pulled aside, while actual criminals will breeze on through.
Re: (Score:3)
The birthday paradox is about comparing everyone with everyone ("Do two people in this room have their birthday at the same day?"). You're talking about the base rate fallacy.
Re: (Score:2)
Depends on how you formulate the problem. If you ask "Is a person one of these 1000" it is no birthday problem. If you ask "Are there two persons with the same identification" it is. But it is a fallacy to think the birthday paradox implies the problem with these 1000. The probably for a match with these 1000 is not affected by the matches between the other people not on the watchlist. If you would want to apply the birthday paradox numbers, you would need to use the bayes formula, which would decrease the
Re: (Score:2)
Current systems are much better, but that doesn't invalidate your main point.
They don't/shouldn't optimize on error rate. These algorithms return a numerical match, so you set can threshold to control the false positive (identify person not in gallery) rate given the galler
Salesweasels exaggerate... (Score:4, Insightful)
...the abilities of current AI
Lazy and technologically clueless people believe the exaggerations.
Current AI is unreliable and should not be trusted for any serious work
At best, it's entertainment, at worst total crap
Re:Salesweasels exaggerate... (Score:4, Interesting)
The average Joe Sixpack goes home and watches CSI or any of these other modern police shows and honestly thinks that this technology is magical. You get crap like this (Castle) [youtube.com]. Or this (CSI: Cyber) [youtube.com]. Or this (also CSI: Cyber) [youtube.com]. I couldn't immediately find the one where "he's moving too fast!" so two people start slamming away at the same keyboard. About the only way for a moderately intelligent human being to watch shit like this if they're about two thirds down a bottle of whisky by the opening credits.
Hmmm. (Score:4, Insightful)
The penalties aught to be a lot higher. I'd say that $300,000 is not enough to cover emotional trauma and the cost of fixing it. Further, US jails are notorious breeding grounds for novel infections, exercise is poor, as is the food, so there's a health impact too. I'd suggest $500,000 would be better, but for that part alone.
But you've also got lost earnings, impact on the family, any debt that resulted, difficulty in finding work later, and emotional impact affecting schooling for the kids, and thus potential loss of earnings across the generations. If you add all of that up, I think an extra $500,000 would be reasonable.
So I'd have given the guy $1,000,000 in compensation.
I'd have fined the police department a further $1,000,000 for malicious wrongdoing, fraudulent charges, and gross misconduct, with that money going into the community.
Those involved should then have been tried and convicted of falsifying evidence, perjury, and misuse of government furnished equipment, with a suitable prison sentence.
Police will take things seriously when the consequences are significant.
need more Bail Reform remove the $ part of bail (Score:2)
need more Bail Reform remove the $ part of bail
Re: (Score:2)
I'd have fined the police department a further $1,000,000 for malicious wrongdoing, fraudulent charges, and gross misconduct, with that money going into the community.
The community (tax-payers) are the ones providing that money.
Re: (Score:3)
Then, they should be demanding lazy, incompetent cops are dismissed. They should be demanding a justice system that prioritizes examining the evidence, not beating a confession out of a 'suspect'. This is precisely why the USA has 'small' government but this technology exposes its second flaw: The government "did something", its usefulness in protecting anybody and preventing further crimes, is not relevant. (The first flaw, 'small' government has no leverage against self-important demagogues.)
This is
Re: Hmmm. (Score:2)
This country has always been fascist for many minorities. We just used to strive to someday be better. We have given up that goal, and now the oligarchy no longer has racial allegiances. All not in the elite are targets.
Re: (Score:2)
Then eliminate qualified immunity and go after the specific and individual arresting, detaining, interrogating, or otherwise harassing officers themselves. It's not exactly rocket science. Liquidate the assets of the involved cops first and foremost, before going after the department's and PoA's and union's operating and benefits funds. Do that enoughes, and the whole: "Do not stop, detain, arrest, definitely not shoot, question, or in any other way darken the day of anyone who has not, in fact*, committ
Re: (Score:2)
Penalties are paid by the tax payers. Criminal charges need to be filed against the investigators and police who willingly failed to follow procedures. Not different in concept, but worse in impact that people charged with fraud.
Re: (Score:2)
Penalties should be paid out of the police union war chest or the pension fund
lol Cops are stupid (Score:3)
None of the smart kids in my AP and Honors classes became cops. It was always the mouthbreathers from the regular and special ed classes who became cops. Of course they don't follow the rules. They're the ones responsible for lowering the standards.
Very small numbers (Score:2)
In the summary, it states
In six cases police failed to check alibis
In two cases police ignored evidence that contradicted their theory
In five cases police failed to collect key pieces of evidence
In three cases police ignored suspects' physical characteristics
In six cases police relied on problematic witness statements
Those are incredibly small numbers. If police make mistakes that seldom, then we're in really good shape.
The reality is that police make many, many more mistakes than are represented by these numbers.
Corroboration (Score:2)
This sort of thing is why the statements of police officers are not accepted by the court as evidence without corroboration. Preferably corroboration from independent sources (e.g., independent-of-police fingerprint or DNA databases ; CCTV from private individuals or businesses ; or even, scraping the barrel, eyewitnesses), but if nothing else, another po