"But that was not before the falsely accused man had his name and image widely shared. The alert sent by Citizen contained a photo and was seen by more than 861,000 people. It read: "Citizen is offering a $30,000 reward to anyone who provides information that leads to the arrest of the arson suspect."
Citizen told the Guardian in a statement it offered the cash reward "without formal coordination with the appropriate agencies".
"Once we realized this error, we immediately retracted the photo and reward offer," it said. "We are actively working to improve our internal processes to ensure this does not occur again. This was a mistake we are taking very seriously.""
"However, according to data on the force's website, 92% (2,297) of those were found to be "false positives".
South Wales police admitted that "no facial recognition system is 100% accurate", but said the technology had led to more than 450 arrests since its introduction. It also said no one had been arrested after an incorrect match."
"Facewatch is being connected to realtime facial recognition systems that can be tied into the CCTVs of its 10,000+ participating retailers. If your face is on the blacklist and the system recognises you (or if your face isn't on the blacklist and you generate a false positive), the shopkeeper/security staff will get an alert when you enter their premises and they can make you leave. "
"Rekognition indicated high confidence that 28 members of the current Congress were known arrestees. It was wrong in every case. The false positives disproportionately targeted racialized members of Congress.
This, finally, has Congress's attention: members of Congress have sent some pointed questions to Amazon about its Rekognition tool and given them a deadline of Aug 20 to respond. They've also requested an immediate meeting with Jeff Bezos to discuss the topic in depth."
"This led to claims that the software is woefully inaccurate; in fact, police had set the threshold for a match at 60%, meaning that faces do not have to be rated as that similar to be flagged up. This minimises the chance of a person of interest slipping through the net, but also makes a lot of false positives inevitable."
"A preliminary study finds that facial recognition algorithms struggle to identify people wearing masks. The study tested 89 commercial facial recognition algorithms, and the best had error rates between 5% and 50% in matching unmasked photos with photos of the same person wearing a digitally-applied mask. Masks both lowered the algorithms' accuracy rates and raised the number of failures to process. The more of the nose is covered by the mask the lower the algorithm's accuracy; however, error rates were generally lower with round masks; and the algorithms generally performed worse with black masks than with surgical blue ones. False positive remained stable or declined a small amount."
"But an analysis by the Office of Information Technology Services issued last month "acknowledges that the risks of the use of (facial recognition technology) in an educational setting may outweigh the benefits."
The report, sought by the Legislature, noted "the potentially higher rate of false positives for people of color, non-binary and transgender people, women, the elderly, and children."
It also cited research from the nonprofit Violence Project that found that 70% of school shooters from 1980 to 2019 were current students. The technology, the report said, "may only offer the appearance of safer schools.""
"But it isn't just major retailers deploying facial recognition software. Backlash to private use of facial recognition culminated on Wednesday when Livonia skating rink in Michigan was accused of banning a Black teenager after its facial recognition software mistakenly implicated her in a brawl.
Lamya Robinson told Fox2 that after her mom dropped her off at the skating rink last Saturday, security guards refused to let her inside, claiming her face had been scanned and the system indicated she was banned after starting a fight in March."
"The combination of the infinite scroll and lack of live links has led to a very specific kind of information economy on the app. Creators whose brand involves discussing news or popular culture (and there are a lot of them) opt to screenshot headlines, summarize articles, and offer their takes. But it's impossible to know whether the creator even read the article or if they're just offering a summary of the online discourse. Even worse are the accounts that spread false headlines, whether they know it or not. Without live links, all these possible situations are presented to you in the exact same way: with a familiar face on your FYP speaking authoritatively."
"It forecast 14 earthquakes within a 200-mile area of the estimated epicenter and also made a very accurate forecast regarding their intensity, a report on the university's website said. It failed to warn of just one earthquake and gave eight false predictions.
The research team trained the AI to detect statistical bumps in real-time seismic data that the research team had paired with previous earthquakes, the report explained. Once trained, the AI monitored for signs of approaching earthquakes.
"Predicting earthquakes is the holy grail," said Sergey Fomel, a professor at UT's Bureau of Economic Geology and a member of the research team, adding: "What we achieved tells us that what we thought was an impossible problem is solvable in principle.""
"Georgetown University's Center on Privacy and Technology highlighted the April 2017 episode in Garbage In, Garbage Out, a report on what it says are flawed practices in law enforcement's use of facial recognition.
The report says security footage of the thief was too pixelated and produced no matches while high-quality images of Harrelson returned several possible matches and led to one arrest."
"Cellebrite, the Israeli-founded and now Japanese-owned company behind some of the software, claims a wider rollout would solve problems over failures to disclose crucial digital evidence that have led to the collapse of a series of rape trials and other prosecutions in the past year. However, the move by police has prompted concerns over privacy and the potential for software to introduce bias into processing of criminal evidence."
"Police are facing calls to halt the use of facial recognition software to search for suspected criminals in public after independent analysis found matches were only correct in a fifth of cases and the system was likely to break human rights laws."
"The tech giant revealed updates to the controversial tool on Monday that include improving the accuracy and functionality of its face analysis features such as identifying gender, emotions and age range."
"The app, which scraped billions of photos from the likes of Facebook, YouTube, Venmo, and other online platforms, drew the world's attention last weekend following a detailed report in the New York Times. The app's supposed capability to identify practically anyone from even low-quality photos frightened privacy advocates and officials. And today, one of the latter - New Jersey's attorney general Gurbir Grewal - actually did something about it."
"But even cleverest AI can be fooled with the simplest of hacks. If you write out the word "iPod" on a sticky label and paste it over the apple, Clip does something odd: it decides, with near certainty, that it is looking at a mid-00s piece of consumer electronics. In another test, pasting dollar signs over a picture of a dog caused it to be recognised as a piggy bank."
"In a recent test of Amazon's facial recognition software, the American Civil Liberties Union of Northern California revealed that it mistook 26 California lawmakers as people arrested for crimes. "
"Game players-like drivers-often have to reach conclusions without full understanding of what the other players-or drivers-are doing. So more researchers are applying game theory to train self-driving cars how to act in uncertain situations."