Louisiana Cops Use Facial Recognition Tech To Arrest The Wrong Person For String Of Robberies
from the fitting-the-description-when-you-don't-fit-the-description dept
This is always going to be a thing with facial recognition. Hundreds of algorithms have been tested. Pretty much every single one does worse “recognizing” minorities than it does recognizing the predominant deployers of facial recognition tech: white males.
In 2020, the Detroit PD managed to rack up consecutive false arrests by relying far too heavily on the conclusions (badly) drawn by its facial recognition tech. This PD will rack up more false arrests in the future, seeing as its tech is wrong nearly 100% of the time.
Cops in New Jersey did the same thing later that year, arresting a man who voluntarily showed up at the station after being surprised to learn there was a warrant out for his arrest. The victim spent 10 days in jail and then the rest of his personal savings clearing his name.
The next headline-making misfire by facial recognition tech takes us down south, where the go-getters at two Louisiana law enforcement agencies combined forces to fuck up.
[A] recent attempt by the Jefferson Parish Sheriff’s Office to nab a high-end purse thief via facial recognition ended badly for a Georgia man who was jailed for almost a week over a false match, his lawyer says.
A detective took the algorithm at face value to secure a warrant to arrest Randal Reid, 28, in the June theft of luxury purses from a Metairie consignment shop, attorney Tommy Calogero said.
A Baton Rouge Police Department detective then adopted JPSO’s identification of Reid to secure an arrest warrant alleging he was among three men involved in another luxury purse theft the same week at a shop on Jefferson Highway, court records show.
This “adoption” of a mismatch made its way into the hands of Georgia law enforcement. Officers in Dekalb County, Georgia pulled over Reid and surprised him with the news he was wanted for robberies committed in a state he’d never visited. He was booked into the county jail on November 25. He was finally released on December 1, after the Jefferson Parish Sheriff’s Office “rescinded” the warrant.
Now, Reid’s lawyer does admit the facial match was pretty close, calling his client a “spitting facial image.” But it’s not just faces that need to match. The rest of the description should match as well. And that’s where things diverge. The suspect captured on store cameras was at least 40 pounds heavier. The suspect also did not have a mole on his face, like Reid does. Unless officers were inclined to believe Reid lost 40 pounds in five months while growing a mole on his face, Reid should have been released immediately.
But that didn’t happen. And it shows yet another problem with relying solely on facial recognition tech to obtain arrest warrants: the only thing this tech looks at is faces. There’s a lot more to a person than the set of physical features residing above the neck.
Then there’s the rest of the ridiculousness surrounding this screw up. Law enforcement officers don’t mind being wrong. That’s just another attack vector for warrantless searches and cash seizures. They just hate it when everybody else finds out that they’re wrong. That’s when they go into lockdown mode, preventing the press from accessing anything that might further expose their careless patterns and practices.
Sheriff Joe Lopinto’s office did not respond to several requests for information on Reid’s arrest and release, the agency’s use of facial recognition or any safeguards around it. That office also denied a formal request for the July 18 arrest warrant for Reid and copies of policies or purchases related to facial recognition, citing an ongoing investigation.
Baton Rouge police also did not respond to questions about its warrant for Reid’s arrest.
When cops score a win, they can’t stop talking about it. When things go badly, they refuse to engage with the same press they rely on to reprint press releases touting their latest triumph over evil. It’s always the same and it never gets any less annoying or stupid.
However, NOLA did manage to obtain something that’s probably going to be useful to Reid and his ongoing lawsuit.
The warrant, signed by 19th Judicial District Judge Eboni Rose, does not say how Lopinto’s office identified Reid.
That looks like a pretty serious omission. It’s going to be pretty difficult for officers to claim they acted in good faith when the warrant was obtained in bad faith. And it may get worse for the JPSO if future document releases show the agency plays fast and loose with facial recognition matches, considering nearly every tech provider warns law enforcement agencies matches should never be used as the sole basis for arrests.
Occasional failures by facial recognition tech shouldn’t just be brushed aside as the inevitable outcome of widespread use. These tools are touted as being better than fallible humans at identifying suspects. But cases like these show the tech is little better than the officers using it, who far too often decide most black men look alike so why not just arrest the next one officers come across. We need tools that make police better, more accurate, and less likely to default to their biases. So far, though, facial recognition tech is just playing into the preconceptions of officers, but allowing them to blame the machine, rather than their own impulses for blown calls.
Filed Under: arrests, baton rouge, baton rouge pd, dekalb county, facial recognition, georgia, jefferson parish, jefferson parish sheriff's office, joe lopinto, louisiana, randal reid, robberies


Comments on “Louisiana Cops Use Facial Recognition Tech To Arrest The Wrong Person For String Of Robberies”
As an old saying goes: “Success has many fathers; failure is an orphan.”
Re:
Another issue is the fact that some people turn their brains off when a computer tells them to think something. If people will drive into a river directly in front of them because a GPS told them it was a road, other will arrest a guy because the computer told them it was the right guy even though their eyes tell them it’s not.
That’s the real dystopia. Not the presence of AI, etc., but because some people with power decide that it means more than their own eyes.
Re: Re:
Computer says no.
Re: Re:
Not just “people”, not just “AI”.
Remember the Appeal that affirmed an officer’s accusation about “weaving in lane” and “crossing the right-hand line onto the shoulder”, despite the officer’s car video that didn’t show any of that?
Finding the evidence you want to believe, then holding onto it despite proof contrary is not just related to computer-sourced information.
Re: Re: Re:
True, but at least that situation indicates good old corruption and deliberate lying. The situation in front of us here indicates that they will be abusing innocent people even if they think they’re acting honestly…
This comment has been flagged by the community. Click here to show it.
Well that’s not true.
Well, no. Presuming it advances as all technology does at some point it will be better at telling people apart than people are.
Don’t get me wrong: I’m utterly against facial recognition and I think it should be banned wholesale. But that’s cuz I don’t want to live in a surveillance state. Fundamentally I’m worried that it will work too well (or at all), not that it doesn’t work well enough.
Re:
Technology like this will always have an error rate. That the error rate will be lowered is no excuse to rely on technology alone as the sole reason for putting someone in jail.
Re: Re:
Not to mention many of the photos taken by police mugshots over the years are too low resolution compared to what’s needed for a high probability of matching even with a good system.
Garbage in, garbage out.
Re:
“Presuming it advances as all technology does at some point it will be better at telling people apart than people are.”
Which is where the danger lies. We already have stories about the wrong people being arrested, detained or otherwise attacked because the computer told them there was a problem even if using their own eyes would tell the cops that the computer was wrong. That’s with a well-known high level of false positives.
Abuses will get worse if the cops think they don’t even have to bother double-checking, so it getting better is actually a vey bad thing.
This comment has been flagged by the community. Click here to show it.
SO that’s an actually dumb thing to say. Everything has an error rate. There will always be an error rate. Combination of factors? Error rate. It is conceivable (not likely) that a single factor could have a smaller error rate than many others together.
There will always be an error rate. If you put people in jail you accept that occasionally you will put innocent people in jail. It WILL happen.
None of which has anything to do with why I find facial recognition abhorrent.
Re:
“None of which has anything to do with why I find facial recognition abhorrent.”
Because it’s all about you.
Why might someone else find use of facial recog to be a very bad idea? Maybe because they were that person the error rate fucked up? Nah … no way! Being caught up in the gears of political bullshit machinery is just another of those things all adults just accept – right?
Re:
SO that’s an actually dumb thing to say. Everything has an error rate.
When you consider the other half of the sentence:
…is no excuse to rely on technology alone as the sole reason for putting someone in jail
I’d be really, really, fucking careful about blindly following the GPS wherever it takes you. The police in this instance put this man in jail despite not matching weight, a unique mole, and (kinda important) a lack of proximity to where the crime happened.
Error rate in these situations can’t factor in willful ignorance of other factors.
If you’re going to color code technology, heres a little cheat sheet:
African (Black, Negro)
European (White, Caucasion)
Australian (Black, Aboriginal)
Asian (Yellow, Mongoloid)
American (Red, Indian)
Indian (Brown, Indian)
Yellow and brown seem to be in a race to 2 billion within borders too. I think I just read India is just beating China in population.
https://infogram.com/race-of-the-world-population-1go502yg18k62jd
You’d think they’d develop their own racial profiling software.
Yay!! We’re gonna see more tax dollars wasted on paying off people who are the victims of cop stupidity rather than holding them accountable for their own screwups.