Did Walmart cross a line with its facial recognition tech test?
Whether or not the public at large will accept the use of facial recognition in their daily lives remains to be seen, but it is clear that retailers are willing to investigate whether the controversial technology has a role to play in stores. Fortune, for example, recently reported that Walmart had been testing the use of facial recognition technology to identify shoplifters.
According to the Fortune article, Walmart used a solution called FaceFirst, which works by scanning the face of every shopper that comes through the door. The software then checks each face against a preexisting database of suspected shoplifters. If the face of a customer matches with one in the database, the store’s staff are alerted on their mobile devices.
The software was in place in stores throughout several states for several months. The use was discontinued due to a failure to produce a positive return on investment.
Fortune attempted to get in touch with various other retailers about the use of FaceFirst, but only Walmart admitted to having used it.
A consumer survey quoted in RetailWire in August of 2015 noted that 75 percent of consumers said they would not shop at a store that used facial recognition for marketing purposes.
![[Image: FaceFirst Retail]](/public/images/discussions/18663/facefirst-still.jpg)
Whether consumers are more comfortable with facial recognition being used for loss prevention remains to be seen, as is whether facial recognition intended for loss prevention purposes will eventually creep into other areas of a business.
Privacy watchdogs have been attempting to set specific guidelines for the use of such potentially invasive technology, as well as similar technologies used on social media. The results have not been what they had hoped.
Longtime online privacy advocate organization the Electronic Frontier Foundation (EFF) had, according to a press release, been engaged in a National Telecommunications Information Administration process to arrive at sensible limits for the use of facial recognition technology in the private sector and the sharing of information collected with government agencies.
According to a statement on EFF.org from June of 2015, the EFF pulled out of the process along with eight other privacy organizations after 16 months due to their perception that companies refused to place the most basic limitations on the use of the technology for the sake of privacy.
- Walmart’s Use of Sci-fi Tech To Spot Shoplifters Raises Privacy Questions – Fortune
- Wal-Mart Stores, Inc. On The Spot Over Use Face Scanning Technology In Stores – Invest Correctly
- Are shoppers ready for their facial recognition close ups? – RetailWire
- EFF and Eight Other Privacy Organizations Back Out of NTIA Face Recognition Multi-Stakeholder Process – EFF.org
Discussion Questions
Do you see a practical use for facial recognition technology in retailers’ loss prevention efforts? Do you expect that privacy advocates to derail the practice?
Poll
BrainTrust
Paula Rosenblum
Co-founder, RSR Research
Adrian Weidmann
Managing Director, StoreStream Metrics, LLC
I suppose truly identifying previous LP offenders is a decent use of facial recognition technology and is “better than racial profiling.” But one question that must be answered is: Is there a statute of limitations? Or will one mistake follow the offender forever?
And what’s to keep retailers from using facial recognition for profiling as well?
I think it’s a DOA technology.
This is the only logical route to go. A video camera is the same thing and those are in every store you know. Facial recognition can be done from video too. For privacy freaks — what if your neighbor has a video camera on their porch? Will you not visit them? What if your parents do?
Retailers need to protect their “homes” from organized crime. They are not using the facial recognition for marketing or selling. They are trying to stop the mafias of shrinkage. Smart move Walmart and let the privacy freaks rant on something else.
Side note: Most Millennials and Generation Zers would not care — they post selfies everywhere!
Apparently not, if they discontinued the test because it didn’t produce enough hits to pay out. It’s not a privacy issue, it’s an ROI issue.
Facial recognition, like biometrics, will continue to be considered as an additional security safeguard against fraud, shoplifting and even VIP shopper identification. The key to mollifying the consumer’s privacy concerns lies in the communication to the shopper as to why these technologies are beneficial to them.
Given the increasing number of credit card fraud cases, stolen identities and other personal information breaches, I believe most consumers understand that it is to their benefit to make sure they are who they say they are when a payment instrument is being used.
In addition, given the increasing number of security cameras both inside and outside retail stores, the majority of shoppers are already aware that they are being watched.
Surely there will be privacy advocates who will find fault with some of the more invasive techniques, but ultimately it will be the consumer who will decide if the measures provide more benefit than inconvenience or intrusion in their personal lives, not an advocacy group.
Sure, it’s a practical use. This technology has done far more good than harm. Without it, the Boston Marathon bombers would still be free and would no doubt have killed more people in more attacks. Many murderers and rapists would never have been caught. Gimme a break. Common sense will, over time, erode the argument of privacy advocates.
Let’s get this straight. If you walk into a store and the owner or clerk recognizes you and greets you personally, this is not offensive. But if a “robot” does the same thing, it’s hellfire and brimstone all around.
I’m sorry that so many people have such a weak understanding of the world they live in that Luddite “privacy advocates” are allowed to continually muck up progress, IF THEY CAN!
I still love Scott McNealy’s to-the-point article, “You have zero privacy anyway, get over it.” (Wired, 1-26-1999.)
As long as shrink continues — and it will, brick & mortar retailers have no choice but to leverage and test all available technologies to limit their losses due to shoplifting. According to NASP, the National Association of Shoplifting Prevention, there are 27 million shoplifters in the US (1 in 11 persons) and 10 million have been caught over the past 5 years. A 2009 global report concluded that retail crime cost a family $435/year.
These economic pressures will force retailers to leverage all available technologies to address this challenging problem. This puts ever increasing pressure on brick & mortar retailers and may become (already is?) a driving force towards reducing physical footprints and inventories as well as even more online sales.
The ROI of FaceFirst implementation of live-feed facial recognition is not an indictment of the overall technology.
Retailers do not need a live-feed analysis and can delay processing several minutes later reducing the costs significantly — in some cases, cloud based processing snapshots with sharing of data between retailers would reduce TCO.
Computer vision technology would be best served tracking “objects” and “positions” to determine heat maps of in-store traffic flow and activity. These cameras can also detect long lines and when a product is out of stock (put a QR on the back shelf and once items are depleted the camera can view the QR code — out of stock).
There are great cases for computer vision (CV) in retail, but live feed facial recognition for petty shoplifters is not one of them.
It’s not the recognition that makes me queasy; it’s the aggregation of the data.
When I’m out in a privately-owned or public space, I have no expectation of privacy whatsoever. When my identity is captured, however, I have an interest in knowing that it will be used responsibly and protected from bad actors. Considering the current state of cyber-security, I have zero confidence that my personal data is safe from breaches.
If a retailer uses facial rec and machine learning to scan me, classify me, figure out my gender or ethnicity, track me in the store, and even compare my behaviors to the resultant purchases, that’s really OK. The underlying intent there is to use sensing to understand behavior and improve customer experience.
If the same retailer uses facial tech to identify me personally and match my identity against a database, that’s not OK. It’s an invitation for abuses of various kinds, and it results in the creation of a data trove that is irresistible to data thieves.
Bottom line: You don’t need my personal identity to catch me stealing. You just need an observation process. It’s OK to use machine learning to make this more effective. It’s OK use the photo as evidence in an individual instance of shoplifting. It’s not OK to build a persistent time-and-location stamped private database of shopper images. Leave that sort of thing to the NSA.
Privacy aside, I don’t see widespread adoption of the technology among retailers unless it becomes much less expensive and more reliable. It’s one thing to use it in a casino, for example, but for most retailers, it will fail a cost/benefit analysis, which seems to be why Walmart didn’t pursue it further.
The issue of profiling will have to be addressed as well as statute of limitations, and how else the pictures may be used. People may not know how their pictures are being used now. However, if consumers find out that they have been used inappropriately, there will be a huge backlash. Retailers will have to weigh benefits against the actual of the system and the potential cost of a backlash from consumers. At this point using this technology is risky versus a questionable payoff.
As some other commenters have noted, there is concern over “who” owns the central database documenting offenders and “how” citizens can manage their status in that database.
In general, the retailers who pioneer this technology will face a mountain of consumer resistance. Given consumer privacy concerns about personal data, I think this technology can wait for another 2-3 years.
To my knowledge, the vast investment that retailers have made in physical security (item tags and CCTV) is effective. I would like to see more in the way of comparison between the existing investment and that required to mount a facial recognition campaign.
Casinos do this to catch cheaters, why not Walmart? The privacy advocates are not protesting in Las Vegas.
Yes indeed. Casinos have used it for many years and Vegas continues to draw folks from around the world. Consumer acceptance depends on how retailers use and secure the information. No one want another marketing spam email. Video has many positive uses in stores, instantly responding to lines at checkouts to shift staff assignments and provide a better shopping experience, to name just one.