May 6, 2026
whyframeshot/Depositphotos.com
Will Increased AI Security Halt the Selective Rollback of Self-Checkouts?
With recent news emerging that Walmart — alongside Target and Costco, among others — is rolling back self-checkouts in certain locations, both the reasons as to why as well as the proposed improvements are hotly debated.
While Walmart itself is stating that customer and associate feedback, as well as a desire to improve the checkout experience, are behind the move, many analysts are suggesting that rampant shoplifting and checkout theft are more likely drivers.
According to Capital One Shopping data from earlier this year, self-checkout represents 40% of grocery store registers nationwide — but also that more than one-quarter (27%) of shoppers (or 36 million-plus Americans) admitted to using self-checkout to steal. A further 55% of these stated intentions to re-offend.
Despite this, the self-checkout market is expected to grow from $2.54 billion in 2025 to more than $5 billion in 2030. Could improved computer-driven, AI-enabled security measures create a more frictionless self-checkout experience, or will retailers simply have to accept the tradeoffs tied to replacing traditional cashiered lanes with self-checkout installs?
Could AI-Driven Computer Vision Help Guide Self-Checkout to Where it Needs To Be?
Bill Miller, president of retail tech provider GK Software USA, highlighted both the common elements of the tug-of-war represented by the fluctuating self-checkout situation, as well as some proposed security solutions.
Writing for Retail TouchPoints, Miller first pointed to the current reality: Despite massive shrink, self-checkout remains “too operationally and economically valuable to abandon.” But despite rollbacks taking place in certain localities, could these be mitigated by an improved security system — one reliant on AI-driven computer vision?
“Computer vision is real-time visual recognition. Cameras positioned at self-checkout lanes capture a continuous feed of every movement and item in the checkout area,” Miller wrote.
“Machine learning models trained on large volumes of transaction video streams analyze that feed to understand human behavior at self-checkout and recognize items. This technology distinguishes a completed scan from a missed one, a legitimate bag placement from an item that bypasses the scanner, or a premium product being rung up as a cheaper alternative,” he added, noting that these models are not reliant on barcodes or weight but rather hand position, item trajectory, and item movement. Deviations are flagged and customers are asked for a re-scan, while associates are immediately notified with a short video clip of the error to provide context.
Miller went further to say that, ideally, these systems would be installed on a per-lane or at least per-store basis, cutting down on latency and reliance on cloud compute to solve problems that could be contained within each retail location.
Additionally, by analyzing patterns at the SKU and lane level (routing findings upstream to merchandising or category management to isolate root causes); auditing escalation logic on a routine basis (to prevent over-alerting and false positives, with the goal of reducing friction or customer annoyance more broadly); and tightening intervention outcomes (assessing signals such as routine dismissal of alerts by associates to see what, exactly, is going wrong); the overall workflow can be tightened to a desired level, per Miller.
“Computer vision stands apart from most AI initiatives because it solves a visible, costly, well-defined problem with results that are measurable at the lane level from the start,” he concluded.
BrainTrust
Anil Patel
Founder & CEO, HotWax Commerce
Georganne Bender
Principal, KIZER & BENDER Speaking
Nolan Wheeler
Founder and CEO, SYNQ
Recent Discussions
Discussion Questions
Can computer-driven, AI-based security improvements help to lessen self-checkout rollbacks, or is there a larger issue at play?
Do you believe that these selective self-checkout rollbacks are driven by a desire to improve CX, by theft or shrink concerns, or something else entirely?
Do you believe an increasing percentage of checkout will be handled by self-checkouts rather than traditional cashiers in the future, regardless of targeted withdrawals?
Poll
Loading …
Reports of the death of self-checkout are greatly exaggerated. While there has been a selective scaling back and increased restrictions, such as item limits, a lot of stores still have self-checkouts. That said, shrink – which is both deliberate and accidental – is still an issue. If AI can monitor and safeguard against this, it will be of huge benefit. However, I still think the best solution is RFID type checkouts, like Uniqlo has, which negate the need for item scanning altogether.
Experiencing technology and the speed of development, I do not doubt that a solution, RFID or something we never thought of, will solve the problem.
The pull-back in SCO is almost entirely shrink related, as customers have come to expect to scan their own items for years now, thus weakening the customer service argument. For CV/AI solutions to actually impact the growing shrink problem, the question is what amount of latency is the general public willing to accept? Cloud-based lag is likely too long, so on-prem servers or edge compute is necessary to keep the lines moving.
Self checkout has a limit in the number of people who want it. It moves from laziness to being offended at taking someone’s job. Maybe if it was like the old self-service gs stations, where you pie a few cents less?
there are a lot of issues. Shrink is the big one. Failure of tech is another.
this is not new news. I’m surprised it keeps coming back as a growing technology
My, my what can’t AI do ?? Or to be more precise, what topic won’t it be gratuitously dragged into ?? (I fear that may be a null set)
The gap between what some sharp-tongued sales rep claims their offering – AI or otherwise – will accomplish, and what it will actually do in the real world is large…vast.
When I can go for more than a few weeks without encountering a self-check that isn’t either unstaffed or malfunctions with every-other item scanned – or both – I’ll put some faith in the premise offered here; I don’t expect it to be soon.
Self-checkout works. We know that. We also know the problems. Now, we need to solve the problems. If we can send people into space (beyond the moon), we can surely fix the self-checkout shrink problems.
I don’t ever see 100% self-checkout for major retailers. There will always be a need for a cashier, or at least someone to support customers when they need help.
Maybe 95%?
Shades of grey in here. Store Associates for safety reasons do not approach/stop shop lifters in motion. Vision tech works well agreed, but now you get to the judgement side. Will AI assess the scenario and instruct the Store Associate to stay away – this looks like a dangerous self scanning cheat……..or please ask this seemingly quiet looking self scanning cheat if they need assistance and we hope that quiet looker is not going to cause a safety issue?
I have no doubt AI, whether through machine learning or other means, via computer vision can improve self-checkout by helping to eliminate shrink, whether intentional or accidental. However, this does nothing to improve the customer experience, which frankly, for most SCO systems is abysmal at best. The best implementations rely on RFID, Uniqlo and Zara come to mind, but these systems tend to find there place in apparel stores versus grocery and home improvement which tend to overly rely on SCO technology.
Another exception I’ve found is Costco, where they have wisely realized that the purpose of new checkout technology is not labor reduction but rather customer experience improvement. Their self checkouts have plenty of staff helping customers and pre-scanning items to speed the line. This works to the point that a typical Costco experience during their busiest times still move customers through the line as if there was no line in the first place. All SCO experiences should aspire to this level of effectiveness, however, the reality is most do not.
I’m not a believer that the driving force behind the pull back of existing SCO systems is shrink related. The customer experience in those systems was a huge negative for those brands, and customers will speak with their feet by refusing to use those systems. I’d watch retailers like Michael’s who seem to think it’s ok to move to a SCO only model in many of their stores and see how foot traffic is affected. In local Michael’s I’ve visited, I see fewer people checking out and often see some walk away rather than be forced into using SCO.
AI helps self-checkout evolve with superior accuracy and specificity for risk mitigation. Immediate video playbacks help store associates promptly resolve issues so they can focus on serving rather than policing.
Self-checkout is a very useful added service in a retailer. Unfortunately, retailers seem, somehow, to wish to get rid of human employees. Thus, they have over-stated how much self-checkout matters and tried to force it ONTO customers. That strategy never works.
If retailers believe their self-checkouts aren’t used enough, it’s probably because they expected far too much of them — the common error in tech from GroupOn to “Just walk out” which are only meaningful for a narrow portion of shoppers or shopping trips.
As to security, that sounds like an AI technology desperately looking for a problem to solve — whether it matters or not.
If the goal of self-checkout is to eliminate human employees, the problems with it will never be solved. If the objective is to give the customer a better experience, the solutions are just around the corner.
I attended a panel discussion today at Shop! MarketPlace on designing stores to reduce theft. It was an interesting conversation about shrink, security, locking up merchandise versus keeping it accessible, smart shelves, and AI technology. The bottom line is that there are plenty of tools that can help deter theft, but there’s no magic fix. The most effective theft deterrent is still people.
I agree, “The most effective theft deterrent is still people.” We have a huge cultural problem in this country regarding ‘a little theft’.
Bill Miller makes a strong operational case for computer vision. But the deeper problem isn’t technology. When 27% of shoppers steal at self-checkout and 55% of those plan to do it again, we have a failure of the social contract. The design flaw is in the retailer’s response, which introduces friction that punishes the honest 73%. The right SCO design makes the honest path effortless, and lets AI work invisibly on behavioral risk signals in the background. No weight sensors penalize reusable bags. No false alarms, treating every shopper as a suspect. Costco figured this out: staff facilitates, not police. A fast, uneventful checkout is the desired experience.
Mo, the numbers astound me. Those numbers suggest it is not financial pressure on shoppers but simply a desire to steal. Is that too harsh?
Gene, it’s complex. Some shoppers figure they deserve a “discount” for having to scan and bag their own groceries. Some surveys show that those earning six figures are twice as likely to deliberately skip scanning items as those making 30K/yr or less. The repeat offenders have rationalized their behavior as non-criminal. The idea of getting away with it proves too tempting for many.
Deserve?
AI-based security improvements will absolutely help lessen some of these selective rollbacks because they change the economics of self-checkout. Early self-checkout models came with an understood tradeoff around shrink, but AI and computer vision give retailers far more visibility and control than they had before.
Self-checkout has been a positive additive for the customer experience, a relief when service checkout lanes are long. But retailers who’ve installed them are naturally making tweaks based on what the total performance picture shows. AI will help identify and fix some of the challenges during checkout, but the numbers will tell the story, and determine the plan. If stores are really having a shrink/theft issue, and customers are walking out the door with product they didn’t buy, service checkouts still remain the best option.
I’ve been in stores when there were empty checkouts and a short line for self-checkout.
And there are shoppers that actually want to interact with staff (and it’s a fair number of them).
Self-checkout is not going away because it helps retailers improve checkout speed, reduce labor pressure, and handle higher transaction volume. Most of these rollbacks are not happening because customers dislike self-checkout. They are happening because retailers are dealing with higher theft and inconsistent execution at the store level.
AI-driven computer vision can help reduce missed scans and improve monitoring, but technology alone will not solve the issue. Self-checkout still needs associates to monitor lanes actively, help customers during checkout, and respond quickly when issues are flagged. The future will likely be a mix of self-checkout and traditional lanes, with retailers adjusting based on customer behavior and store formats.
Forgive me for going off topic.
I have spent considerable time living in China, multiple times since 2016. This would be an unheard-of discussion. Theft would not be an issue. It isn’t part of the culture. The comparison is a sad statement for American culture.
And, be assured, it isn’t because of potential punishment.
Yes, AI will be helpful. How could it not be? But…for me this will boil down to how the words ‘always’ and ‘never’ get used in a conversation about any given retailer. I buy a lot of stuff at Home Depot. Their self checkout always works smoothly and I never have a problem. There are always 2 humans manning every 6 self checkout stations. They are always friendly and helpful. Meaning, “I’m standing right here.” It’s smooth sailing every time…always. Conversely, at my local grocery store, self checkout is always a hassle, especially if I have fresh veggies and fruit, which is 97% of the time. There is one human handling 8 self checkout stations, and the flashing light always seems to be pinging at one station or another. We all stand around waiting while the human bounces from station to station. It’s never hassle free…never.
So we have a ways to go in solving the convenience/friction/efficiency equation in retail.checkout. Someday we may even get to a ‘just walk out’ system that works for both the customer and the retailer. Honest humans and a little technology would make a great combination.
Self checkout is preferred by some customers, and also by thieves. The general issue is that many retailers view self checkout as a labor saving device and it marginally added to the customer experience for selected shopping groups. Increased shrink as a result as theives are emboldened to scan and run without worrying about getting past group of cashiers and baggers at checkout. AI maybe improve the identification but do retailers have enough front end staff to do something about it.
Personally I think shrink went up before some thieves feel more comfortable cheating a machine over a cashier line.
It has been interesting to watch the comments in this discussion grow. As someone working in this space, I wanted to share that just this last year alone, our technology recovered over $500M of product that would have otherwise “walked-out” of the store. That is real value, impact felt at our customers’ P&L.
It is proven over and over that this is all about human-centric AI design. AI-powered Computer Vision – Vision AI – won’t solve for the total complexities of the checkout process. Retail has offered one of the first real-world examples of Physical AI in action. Bravo to all retailers “who picked a lane – and embarked on this journey”. It is not a simple “intelligent overlay on top of existing systems of record”. We are learning daily how to integrate this “automated teller machine” into our daily activities.
The responsibility of a successful implementation is equally shared by the retailers, the technology providers, the shopper, and the associates. It truly takes a village to drive the ROI in both the effectiveness and experience of this solution,
More Discussions
What Are the Next Steps for Ross Stores As Sales Soar and Expansion Plans Continue Apace
Ross Stores just delivered a very impressive first-quarter report card ending May 2, as Retail TouchPoints…What’s Behind the ‘Revenge Shopping’ Taking Place, Despite an ‘Unaffordable’ Economy?
What are the primary motivators behind consumers remaining locked in on “revenge shopping”?What Does a Michaels Push Into Party Supplies Mean for the Staples and Party City Team-Up?
With earlier coverage and discussion covering last month’s news that Staples would be expanding its Party…
Reports of the death of self-checkout are greatly exaggerated. While there has been a selective scaling back and increased restrictions, such as item limits, a lot of stores still have self-checkouts. That said, shrink – which is both deliberate and accidental – is still an issue. If AI can monitor and safeguard against this, it will be of huge benefit. However, I still think the best solution is RFID type checkouts, like Uniqlo has, which negate the need for item scanning altogether.
Experiencing technology and the speed of development, I do not doubt that a solution, RFID or something we never thought of, will solve the problem.
The pull-back in SCO is almost entirely shrink related, as customers have come to expect to scan their own items for years now, thus weakening the customer service argument. For CV/AI solutions to actually impact the growing shrink problem, the question is what amount of latency is the general public willing to accept? Cloud-based lag is likely too long, so on-prem servers or edge compute is necessary to keep the lines moving.
Self checkout has a limit in the number of people who want it. It moves from laziness to being offended at taking someone’s job. Maybe if it was like the old self-service gs stations, where you pie a few cents less?
there are a lot of issues. Shrink is the big one. Failure of tech is another.
this is not new news. I’m surprised it keeps coming back as a growing technology
My, my what can’t AI do ?? Or to be more precise, what topic won’t it be gratuitously dragged into ?? (I fear that may be a null set)
The gap between what some sharp-tongued sales rep claims their offering – AI or otherwise – will accomplish, and what it will actually do in the real world is large…vast.
When I can go for more than a few weeks without encountering a self-check that isn’t either unstaffed or malfunctions with every-other item scanned – or both – I’ll put some faith in the premise offered here; I don’t expect it to be soon.
Self-checkout works. We know that. We also know the problems. Now, we need to solve the problems. If we can send people into space (beyond the moon), we can surely fix the self-checkout shrink problems.
I don’t ever see 100% self-checkout for major retailers. There will always be a need for a cashier, or at least someone to support customers when they need help.
Maybe 95%?
Shades of grey in here. Store Associates for safety reasons do not approach/stop shop lifters in motion. Vision tech works well agreed, but now you get to the judgement side. Will AI assess the scenario and instruct the Store Associate to stay away – this looks like a dangerous self scanning cheat……..or please ask this seemingly quiet looking self scanning cheat if they need assistance and we hope that quiet looker is not going to cause a safety issue?
I have no doubt AI, whether through machine learning or other means, via computer vision can improve self-checkout by helping to eliminate shrink, whether intentional or accidental. However, this does nothing to improve the customer experience, which frankly, for most SCO systems is abysmal at best. The best implementations rely on RFID, Uniqlo and Zara come to mind, but these systems tend to find there place in apparel stores versus grocery and home improvement which tend to overly rely on SCO technology.
Another exception I’ve found is Costco, where they have wisely realized that the purpose of new checkout technology is not labor reduction but rather customer experience improvement. Their self checkouts have plenty of staff helping customers and pre-scanning items to speed the line. This works to the point that a typical Costco experience during their busiest times still move customers through the line as if there was no line in the first place. All SCO experiences should aspire to this level of effectiveness, however, the reality is most do not.
I’m not a believer that the driving force behind the pull back of existing SCO systems is shrink related. The customer experience in those systems was a huge negative for those brands, and customers will speak with their feet by refusing to use those systems. I’d watch retailers like Michael’s who seem to think it’s ok to move to a SCO only model in many of their stores and see how foot traffic is affected. In local Michael’s I’ve visited, I see fewer people checking out and often see some walk away rather than be forced into using SCO.
AI helps self-checkout evolve with superior accuracy and specificity for risk mitigation. Immediate video playbacks help store associates promptly resolve issues so they can focus on serving rather than policing.
Self-checkout is a very useful added service in a retailer. Unfortunately, retailers seem, somehow, to wish to get rid of human employees. Thus, they have over-stated how much self-checkout matters and tried to force it ONTO customers. That strategy never works.
If retailers believe their self-checkouts aren’t used enough, it’s probably because they expected far too much of them — the common error in tech from GroupOn to “Just walk out” which are only meaningful for a narrow portion of shoppers or shopping trips.
As to security, that sounds like an AI technology desperately looking for a problem to solve — whether it matters or not.
If the goal of self-checkout is to eliminate human employees, the problems with it will never be solved. If the objective is to give the customer a better experience, the solutions are just around the corner.
I attended a panel discussion today at Shop! MarketPlace on designing stores to reduce theft. It was an interesting conversation about shrink, security, locking up merchandise versus keeping it accessible, smart shelves, and AI technology. The bottom line is that there are plenty of tools that can help deter theft, but there’s no magic fix. The most effective theft deterrent is still people.
I agree, “The most effective theft deterrent is still people.” We have a huge cultural problem in this country regarding ‘a little theft’.
Bill Miller makes a strong operational case for computer vision. But the deeper problem isn’t technology. When 27% of shoppers steal at self-checkout and 55% of those plan to do it again, we have a failure of the social contract. The design flaw is in the retailer’s response, which introduces friction that punishes the honest 73%. The right SCO design makes the honest path effortless, and lets AI work invisibly on behavioral risk signals in the background. No weight sensors penalize reusable bags. No false alarms, treating every shopper as a suspect. Costco figured this out: staff facilitates, not police. A fast, uneventful checkout is the desired experience.
Mo, the numbers astound me. Those numbers suggest it is not financial pressure on shoppers but simply a desire to steal. Is that too harsh?
Gene, it’s complex. Some shoppers figure they deserve a “discount” for having to scan and bag their own groceries. Some surveys show that those earning six figures are twice as likely to deliberately skip scanning items as those making 30K/yr or less. The repeat offenders have rationalized their behavior as non-criminal. The idea of getting away with it proves too tempting for many.
Deserve?
AI-based security improvements will absolutely help lessen some of these selective rollbacks because they change the economics of self-checkout. Early self-checkout models came with an understood tradeoff around shrink, but AI and computer vision give retailers far more visibility and control than they had before.
Self-checkout has been a positive additive for the customer experience, a relief when service checkout lanes are long. But retailers who’ve installed them are naturally making tweaks based on what the total performance picture shows. AI will help identify and fix some of the challenges during checkout, but the numbers will tell the story, and determine the plan. If stores are really having a shrink/theft issue, and customers are walking out the door with product they didn’t buy, service checkouts still remain the best option.
I’ve been in stores when there were empty checkouts and a short line for self-checkout.
And there are shoppers that actually want to interact with staff (and it’s a fair number of them).
Self-checkout is not going away because it helps retailers improve checkout speed, reduce labor pressure, and handle higher transaction volume. Most of these rollbacks are not happening because customers dislike self-checkout. They are happening because retailers are dealing with higher theft and inconsistent execution at the store level.
AI-driven computer vision can help reduce missed scans and improve monitoring, but technology alone will not solve the issue. Self-checkout still needs associates to monitor lanes actively, help customers during checkout, and respond quickly when issues are flagged. The future will likely be a mix of self-checkout and traditional lanes, with retailers adjusting based on customer behavior and store formats.
Forgive me for going off topic.
I have spent considerable time living in China, multiple times since 2016. This would be an unheard-of discussion. Theft would not be an issue. It isn’t part of the culture. The comparison is a sad statement for American culture.
And, be assured, it isn’t because of potential punishment.
Yes, AI will be helpful. How could it not be? But…for me this will boil down to how the words ‘always’ and ‘never’ get used in a conversation about any given retailer. I buy a lot of stuff at Home Depot. Their self checkout always works smoothly and I never have a problem. There are always 2 humans manning every 6 self checkout stations. They are always friendly and helpful. Meaning, “I’m standing right here.” It’s smooth sailing every time…always. Conversely, at my local grocery store, self checkout is always a hassle, especially if I have fresh veggies and fruit, which is 97% of the time. There is one human handling 8 self checkout stations, and the flashing light always seems to be pinging at one station or another. We all stand around waiting while the human bounces from station to station. It’s never hassle free…never.
So we have a ways to go in solving the convenience/friction/efficiency equation in retail.checkout. Someday we may even get to a ‘just walk out’ system that works for both the customer and the retailer. Honest humans and a little technology would make a great combination.
Self checkout is preferred by some customers, and also by thieves. The general issue is that many retailers view self checkout as a labor saving device and it marginally added to the customer experience for selected shopping groups. Increased shrink as a result as theives are emboldened to scan and run without worrying about getting past group of cashiers and baggers at checkout. AI maybe improve the identification but do retailers have enough front end staff to do something about it.
Personally I think shrink went up before some thieves feel more comfortable cheating a machine over a cashier line.
It has been interesting to watch the comments in this discussion grow. As someone working in this space, I wanted to share that just this last year alone, our technology recovered over $500M of product that would have otherwise “walked-out” of the store. That is real value, impact felt at our customers’ P&L.
It is proven over and over that this is all about human-centric AI design. AI-powered Computer Vision – Vision AI – won’t solve for the total complexities of the checkout process. Retail has offered one of the first real-world examples of Physical AI in action. Bravo to all retailers “who picked a lane – and embarked on this journey”. It is not a simple “intelligent overlay on top of existing systems of record”. We are learning daily how to integrate this “automated teller machine” into our daily activities.
The responsibility of a successful implementation is equally shared by the retailers, the technology providers, the shopper, and the associates. It truly takes a village to drive the ROI in both the effectiveness and experience of this solution,