Recently, a dozen eggs at a single Safeway in Washington, D.C. were being sold at five different prices to five different customers at the exact same time. So, we're talking same store, same shelf, same eggs. The only difference was who was buying them, and the price gap was not exactly pennies. It was 23%. But groceries aren't the only place this is happening. A JetBlue customer recently posted on X that his flight had jumped $230 in a single day. He was trying to book travel to a funeral. A JetBlue social media representative responded publicly with a suggestion, "Try clearing your cache and cookies or booking with an incognito window." Think about what this sentence actually means. A customer service representative speaking on behalf of a major airline told a customer in writing, in public, that the price of his flight was being influenced by his browsing data. That the same seat on the same plane on the same day could cost less if the airline didn't know he'd already been looking at it. JetBlue deleted the post. They told Gizmodo the reply was incorrect and apologized for the error. They told MarketWatch that fares are not determined by cache data or other personal information. Four days later, a class action lawsuit was filed against JetBlue in the Eastern District of New York. A company told you how to cheat their own pricing system, then spend the next week pretending they hadn't. My name is Al. I have a PhD in computer science, and I analyze AI developments to understand what's actually happening beneath the hype. In this video, I'm going to show you how companies are already charging different people different prices for the same product at the same time. Why the technology being installed in every Walmart in the US this year makes that easier than ever. And what happens if this mechanism reaches the things we generally cannot live without, our shelter, our water, our medicine? Because the infrastructure being built right now does not distinguish between a luxury you can skip and the food you need to survive. It has a regulator-shaped hole where a stop button should be.

In December 2025, Consumer Reports in collaboration with Groundwork Collaborative and More Perfect Union published one of the most significant consumer investigations in recent memory. They've recruited over 400 volunteers across the United States to simultaneously place grocery orders through Instacart from the same stores at the same time for the same products. What they found was that nearly 75% of grocery items were being offered to different customers at different prices. The differences were not trivial either. On some products, the price variation reached 23%. At a single Safeway in Washington, D.C., one dozen eggs were priced at five different levels, ranging from $3.99 to $4.79, shown to different users at the exact same time. Across a typical family's annual grocery spending, the investigation estimated that Instacart's pricing algorithm could add approximately $1,200 per year to some customers' bills. Instacart described these as randomized tests. Five different prices for the same eggs in the same store at the same time. Randomized. The backlash was, of course, immediate. Instacart announced it would halt the pricing experiments. The FTC opened an investigation. The New York Attorney General demanded answers, pointing out that Instacart's under the state's new algorithmic pricing disclosure act were buried on a page only accessible by clicking on fine print text. The California Attorney General launched a separate investigation, too. The House Oversight Committee, in a rare bipartisan move, requested documents from Booking.com, Expedia, Instacart, Lyft, and Uber.

Now, before I go any further, there is a distinction that most of the coverage of this story is collapsing, and it's important to separate the two because they are not the same thing. Dynamic pricing is when prices change based on supply and demand. Airlines have done this for decades. A flight on Christmas Eve costs more than a flight on a Tuesday in February. A hotel room during a festival costs more than the same room on a quiet Wednesday. You can dislike it as a consumer. Everyone does when they're on the wrong end of it, but the mechanism is at least legible. You know why the flight costs more. Demand is high, supply is limited. The price reflects the relationship between the two. Surveillance pricing is categorically different. With surveillance pricing, the price of a product changes for you specifically because an algorithm analyze your personal data and determine what you're individually willing to pay. The input is not supply and demand. The input is you. Your browsing history, your purchase patterns, your location, your behavior, and you cannot see the mechanism, cannot verify it, and in most jurisdictions can't opt out of it. These two are not the same thing, but collapsing both onto one conversation is extremely useful for companies that practice surveillance pricing because they can point at dynamic pricing, which markets have accepted for decades, and basically say, "This is normal. We've always done this." Adjusting the price of a flight based on how many seats are left is market pricing. Adjusting the price of eggs based on what the algorithm infers about your own willingness to pay is something else entirely, and calling both of them dynamic prices is not a clarification. It is just simple camouflage.

With that distinction in mind, I want to talk about what is currently being installed in grocery stores across the United States. Walmart is rolling out electronic shelf labels, digital price tags, to every one of its approximately 5,200 US stores by the end of 2026. The old paper price tags, changed by hand by employees walking the aisles, are now being replaced by centrally managed digital screens that can update the price of every product in every store simultaneously, instantly, from a single system. On the surface, this is an operational efficiency upgrade, and there is a genuine case for it, yes. A Walmart team leader in Ohio said the digital labels cut the time she spent on pricing duties by 75% freeing her up to help customers. When you have a 1,200,000 products per store and price changes on thousands of items per week, eliminating manual tag replacement is not an unreasonable business decision. I made a separate video about what happens when companies replace their workforce with AI and how that frequently backfires. If you haven't seen it, I'm going to link it in the end. But, the efficiency argument is not the whole story. In January 2026, Walmart was awarded a US patent for a system and method for dynamically and automatically updating item prices. A second patent, awarded very shortly after, covers demand forecasting technology designed to estimate what customers will buy and recommend pricing accordingly. Walmart has stated publicly and repeatedly that it does not implement surge pricing, that price updates are people led, that a human associate reviews and pushes approved changes through a secure system, typically outside of shopping hours, so prices remain stable and consistent during the day, that prices are the same for all customers in any given store. Now, I have no reason to doubt that this is Walmart's current practice. What I would observe, however, is that they have now installed the hardware in every store, patented the software to automate pricing decisions, and built the infrastructure to change prices on 120,000 products simultaneously, while assuring the public that they will not use that infrastructure for the thing it was architecturally designed to do. As Matt Hamrick, who is a grocery industry consultant at Alex Partners put it, "Dynamic pricing or anything that smells like it is playing with fire. There is an element of consumer trust being eroded because they don't know that they're getting the best price at any moment."

[clears throat] That assurance is doing a remarkable amount of work given the circumstances, and Walmart is not alone in this. Kroger is experimenting with electronic shelf labels, updating digital tags when online prices change or for weekly promotions. Whole Foods and Aldi are among early adopters testing this technology. This is not one company making an unusual decision. This is an industry-wide infrastructure rollout. So, what data actually powers surveillance pricing? Because this is where the conversation moves from uncomfortable to generally alarming. Noa Gafni Seracusa, a professor who studies algorithmic pricing, explained it in an interview with Harvard Law Today. "Surveillance pricing doesn't just use your shopping history. It can draw on your search history, your email patterns, what you watch on Netflix or YouTube, your app usage, your location data, and the behavioral patterns your devices reveal about your daily life. The algorithm doesn't need to know your name. It needs to know your patterns, and from your patterns, it can infer your income bracket, your urgency, your willingness to pay, and critically, your inability to walk away. This is not a hypothetical capability. There has been an accusation, denied by the company, that Uber has used phone battery level as a pricing signal. The logic being that somebody whose phone is about to die is more desperate to book a ride and is therefore willing to pay more. Your battery percentage is now a negotiating position. 10% charge and it's raining, the algorithm knows you'll just pay anything. The public is not unaware of this. A recent survey by GBIAO Strategies found that 68% of Americans are worried that surveillance pricing will increase the cost of goods. Only 5% believe it will lead to lower prices. 50% say digital price tags would make them less likely to shop in a store, and 67% support banning electronic shelf labels outright. These numbers are extraordinary. 2/3 of the population in the US wants this technology banned, not regulated, not disclosed, just outright banned. In a country that rarely agrees on much, that level of consensus tells you something about how viscerally people understand the threat, even if they don't know technical details on this one.

And here is where the discrimination problem becomes, of course, unavoidable. If an algorithm sets prices based on behavioral data, browsing patterns, purchase history, location, device type, stuff like that, that data inevitably correlates with race, income, neighborhood, age, and disability status, not because anyone programmed it to discriminate, but because human behavior is shaped by socioeconomic circumstances, and an algorithm trained on that behavior will reproduce those correlations in its pricing. A person shopping from a lower income postcode on an older device with a browsing history that signals financial stress will be treated differently by the algorithm than a person shopping from an affluent area on a new phone, not because a human decided to charge them differently, because the data made that decision and nobody really audited it. So, different people, same store, same product, same time, different prices, not because of supply and demand, but because of who you are. This doesn't sound like personalization. That is discrimination with a better user interface. Now, I want to take everything I've just shown you and ask a question that I think most coverage of this topic is not actually asking. And I want to be very clear that what follows is a thought experiment, an extrapolation based on the mechanisms that already exist. Some things are nice to have. You do not need a 17th coat. You probably don't need a third pair of trainers. If an algorithm charges you slightly more for a luxury item because it knows you can afford it, that is annoying. It's arguably unfair, but it's not a survival issue. But some things are essential. Shelter, water, food, energy, medicine. These are what economists call inelastic goods. Goods you must buy regardless of the price because the alternative is not I'll go without. The alternative is homelessness, hunger, or death. If the surveillance pricing mechanism works on eggs, and the Instacart investigation proved that it does, the question then becomes what stops it from being applied to rent, to energy bills, to prescription medication? And the answer as far as I can personally determine is nothing except regulation. The algorithm doesn't know the difference between a luxury and a necessity. It doesn't care. It optimizes for one thing, finding the maximum price you're willing to pay and charging you that.

Now, extrapolate to the logical endpoint, an AI system with perfect information about every customer. It knows your income, your monthly outgoings, your fixed costs, your debts. It knows what you need to survive. In that world, the algorithm doesn't just know what you're willing to pay. It knows what you're able to pay. And the economically optimal strategy from the seller's perspective is to price every essential good at the absolute maximum you can bear. If the algorithm knows you have 5,000 pounds a month to work with, which is generous I know, and it knows that shelter is your most critical need, there is nothing in the logic of the system that prevents it from pricing your rent at 4,900. Not because that is what the market demands, but because that's what you can survive paying, just barely, with almost nothing left. Your food, your energy, your medicine, all priced the same way. Not by what they cost to provide, but what you can be made to pay. This actually has a name in economics. It's called perfect price discrimination. It means the seller captures all consumer surplus, Every penny you would have saved by paying less than your maximum is taken by the seller instead. In textbooks, this is described as the most efficient outcome for the producer and the worst possible outcome for the buyer. In practice, it has never been achievable at scale because no seller has ever had perfect information about every buyer's circumstances. AI changes that. Not today, not completely, but the trajectory is clear and the infrastructure is being built right now. The data collection is accelerating, the algorithms are ever improving, and the companies deploying this technology have not demonstrated any internal mechanism that would prevent them from moving towards this endpoint. Only external pressure in the form of regulation, public backlash, and legal action. And this is not entirely hypothetical. In August 2024, the US Department of Justice filed an antitrust lawsuit against RealPage, a Texas-based property management software company. RealPage provided algorithmic pricing software to landlords across the country, covered approximately 60 million rental units. The DOJ alleged that the software collected confidential data from competing landlords, rents, vacancy rates, lease terms, and used it to generate pricing recommendations that effectively coordinated rent increases across properties that were supposed to be competing with each other. One landlord told RealPage that within a week of adopting the software, they started increasing rents. Within 11 months, rents had risen more than 25%. The DOJ's antitrust chief put it plainly, RealPage was replacing competition with coordination and renters paid the price. RealPage settled in November 2025 without admitting wrongdoing, and the DOJ subsequently sued six of the largest landlords in the country for their role in the scheme. So, this is not about eggs, this is shelter, the most inelastic need there is, and the mechanism, algorithmic pricing, used aggregated data to extract the maximum possible price, is structurally identical to what Instacart was doing with groceries. The same logic, the same infrastructure, applied to the thing you cannot live without.

If surveillance pricing on groceries is not regulated now, while the public is paying attention and the political will exist, the precedent is set for the same mechanism to be applied or to continue being applied to housing, energy, healthcare, and every other essential that people cannot refuse to buy. People would not be choosing between brands, they would be choosing between eating and having a roof. That is not a market, that is a hostage situation with a receipt. There is, however, something generally encouraging happening in the regulatory landscape, and it's worth paying attention to because it's moving faster than most people realize. In April 2026, Maryland became the first US state to pass a law banning surveillance pricing in grocery stores. The Protection from Predatory Pricing Act prohibits retailers and delivery apps from using a customer's personal data, that would be browsing history, location, shopping behavior, to charge different prices for the same item. It also requires prices to stay consistent for at least one business day, preventing rapid algorithmic changes. The law takes effect October 1st, 2026. Critics note it already has loopholes, it only applies to grocery stores and still allows loyalty programs, which can produce similar outcomes just less transparently, but it is the first law of this kind, so it's a first step. New York's algorithmic pricing disclosure act, which took effect in November 2025, takes a slightly different approach. It requires companies to display a disclosure near algorithmically set prices. This price was set by an algorithm using your personal data. The New York Attorney General's office has already challenged Instacart's compliance, arguing that the company buried its disclosures in fine print. Imagine walking into a shop and seeing that sentence printed next to the price of your milk. I mean, the fact that a law was needed to require that disclosure tells you everything about how transparently this technology was being deployed before. At the federal level, Senator Ben Ray Luján of New Mexico has introduced a Stop Price Gouging in Grocery Stores Act, which would ban electronic shelf labels outright in any grocery store over 10,000 square feet. Given that Walmart Supercenters can approach 200,000 square feet, and even the smaller neighborhood market formats exceeds 10,000, such a law would effectively halt the digital shelf label rollout across most of American grocery retail. The House Oversight Committee has launched a bipartisan investigation into algorithmic pricing, requesting documents and data from booking.com, Expedia, Instacart, Lyft, and Uber. The FTC has an open investigation into Instacart's pricing practices. At least a dozen additional states are considering their own legislation now. This is where the energy belongs, I think. As I've said in previous videos, the path through the AI era runs through regulation. AI as a technology is extraordinarily powerful. It can solve problems that exceed human cognitive capacity. It can accelerate research, model complex systems, and push the boundaries of what's scientifically possible. I genuinely believe that. But this particular deployment of AI, using machine learning to determine the maximum price an individual human being can be charged for the food they eat, is not advancing humanity. It's not solving complex scientific problems. It's not supercharging human capability or anything. It is extracting money from people who are already stretched thin in an economy where consumer sentiment has hit a record low and wages have fallen behind inflation for the first time since 2023. The technology has never been the problem, but the deployment is. And the deployment is a choice. Companies choose to build this infrastructure, regulators can choose the terms on which it operates. That is what regulation is for.

And I want to be clear here, I am not saying that all uses of AI in pricing are inherently predatory. There are legitimate operational benefits to digital price management. There are genuine efficiencies in being able to update prices without sending somebody down every single aisle with a label printer. The question is not whether technology can improve retail operations. The question is instead whether the same infrastructure that improves operations also enables exploitation and whether anyone has built a safeguard to prevent the second from following the first. Right now, the answer to the first question is yes and the answer to the second is it depends on where you live and it depends on how fast your legislators are moving. On April 18th, a JetBlue representative typed six words that a multi-billion dollar airline spent the next week trying to retract, clear your cache and cookies. The post was deleted, the apology was issued, the denial was recorded for the lawsuit, but the sentence was already in the world and the thing about a sentence like that is it doesn't just tell you what one airline is doing, it tells you what's possible. It tells you that the infrastructure exists, that the data is being collected and that somewhere between the algorithm and the price tag, there is a system making decisions about what you pay based on what it knows about you. The algorithm is already watching, my friends. The question is whether anyone is watching the algorithm and right now for most people in most places, the answer is not yet. But that is changing state by state, investigation by investigation, lawsuit by lawsuit. The question is whether it changes fast enough. But the cost of buying things is only half of the equation. What happens when companies don't just adjust prices with AI, but remove their entire workforce and let AI run the operation? The results are not what a brochure promised. That's the video that I would watch next. Thanks so much for watching this one. Subscribe and I'll see you all on the next one.