A Baby Thermometer and a New Reading of the Law
In January 2025, the Federal Trade Commission published the first findings of its study into how companies set prices using consumer data. One example stuck with people who read it.
A shopper profiled as a new parent searches for a baby thermometer. The store, reading that profile, pushes the higher priced models to the top of the results. Same product, different shopper, different price. The FTC built that picture from documents it pulled from six intermediaries, among them Mastercard, Accenture, PROS, Bloomreach, Revionics, and McKinsey. Those firms worked with at least 250 retail clients.
That study lit the fuse. By June 2026, three states had passed bans, two dozen more had bills moving, and the trade press settled on one headline: AI pricing is under attack.
That reading is wrong. Lawmakers and judges drew a sharp, consistent line. The target is pricing that reacts to the person. Pricing that reacts to the market, built on your own data and public data, came out protected. If anything, the crackdown raised the value of clean competitive price intelligence, because it is the legally safer substitute for the personal-data method now turning radioactive.
Two Words People Keep Confusing
Most of the confusion comes from treating two different practices as one. They are not.
Surveillance pricing (also called personalized pricing) sets a price from data about the individual shopper. Browsing history, inferred income, device type, location, loyalty tenure, the contents of an abandoned cart. The price moves because of who you are.
Dynamic pricing (also called competitive or market-based pricing) sets a price from market signals. Inventory levels, demand, seasonality, the prices competitors are charging in public. The price moves because of what the market is doing.
A Holland and Knight analysis put the regulatory split plainly. Dynamic pricing "responds to market conditions (inventory, demand, seasonality)," while surveillance pricing "responds to characteristics of the individual consumer rather than the market as a whole."
The dividing line is not whether your price varies. Prices have always varied. The line is why it varies.
What the New Laws Actually Ban
Read the statutes and the pattern is hard to miss. Each one aims at personal data and waves the market through.
Maryland went first. Governor Wes Moore signed the Protection From Predatory Pricing Act on April 28, 2026. It stops large food retailers from using a customer's personal data, location, browsing history, purchase history, demographics, to set personalized prices. It takes effect October 1, 2026. It carves out loyalty programs, promotional offers, and temporary discounts open to any shopper.
Connecticut followed. Its legislature passed HB 5563 in early May, and Governor Ned Lamont signed it in early June, making Connecticut the second state to act. The ban covers customized prices set from personal data. It carves out the familiar discounts and adds a disclosure rule for price-setting devices that use personal data.
New York is the third. On June 10, 2026, the legislature passed the One Fair Price Act (S.8623B and A.9349B). It now sits with Governor Kathy Hochul. The bill bans the use of personal data to "individually tailor or inflate prices," and it expressly preserves coupons, loyalty programs, subscribe-and-save discounts, and senior and veteran pricing.
Attorney General Letitia James, who championed the bill, framed the goal directly. "When this bill becomes law, shoppers will be able to trust that the price they are paying is a fair price, not one dictated by their web browsing history, income, race, or zip code."
Note the words. Browsing history, income, race, zip code. Personal attributes. Not a single mention of competitor prices, demand, or inventory.
The Tell: New York Protects Dynamic Pricing in Writing
The clearest proof sits inside the New York bill. It does not just leave dynamic pricing alone. It writes the protection down.
The One Fair Price Act bans personalized pricing outright. For dynamic pricing, the kind that reads market conditions and not the shopper, it keeps the practice legal and adds only a transparency step. Say a system changes a price more than once in a 24-hour window. The seller has to disclose that dynamic pricing is in use, how often prices change, and what drives the changes.
Sit with that contrast. One method is banned. The other is permitted, with a label. The label is not a punishment. It is the state telling shoppers that the price reflects the market, not a file on them.
This also clears up an earlier law. New York's Algorithmic Pricing Disclosure Act took effect November 10, 2025. It requires a notice reading, "THIS PRICE WAS SET BY AN ALGORITHM USING YOUR PERSONAL DATA," with penalties up to 1,000 dollars per violation. That label only triggers when a price uses the specific consumer's personal data. Pricing that varies on supply, demand, or general market conditions does not trigger it at all.
California Draws the Same Line From Two Directions
California is worth a closer look because it hits the issue from both the consumer side and the antitrust side, and both point the same way.
On the consumer side, AB 2564 passed the Assembly on May 27, 2026. It defines surveillance pricing as a customized price for a specific consumer based on "personally identifiable information collected through electronic surveillance technology." Penalties run up to 12,500 dollars per violation, tripled for intentional conduct. The Attorney General, district attorneys, and large-city attorneys can sue for penalties, and consumers can sue for injunctive relief. It carves out cost-based differences, publicly disclosed eligibility discounts, and loyalty programs open to all.
On the antitrust side, AB 325 took effect January 1, 2026. It restricts "common pricing algorithms," tools used by two or more firms that price using competitor data. The concern there is collusion, not personalization. A single firm using its own data is outside the rule. This is the one nuance brands should hold onto. Sharing a tool that pools rivals' nonpublic data is a different and older legal problem. Reading prices a rival already posted in public is not.
California also showed it will enforce. On Data Privacy Day, January 27, 2026, Attorney General Rob Bonta opened an investigative sweep. He sent inquiry letters to grocers, hotels, and online retailers, asking whether they use consumer personal information to set prices, and probing their price experiments.
The Courts Agree, and They Reward Clean Data
Legislatures are one signal. Judges are another, and they have been clearer still about which data is safe.
The sharpest example is Mach v. Yardi Systems. In an October 2025 ruling, a California state court granted summary judgment to Yardi in an algorithmic pricing case. The reason is the whole point of this article. The court found that Yardi's Revenue IQ software "does not, and by design cannot, use any client's confidential pricing information to recommend pricing for any other client." Landlords entered their own data for their own use. No rival's nonpublic data flowed into anyone else's recommendation. No commingling, no antitrust violation.
The contrast case is RealPage. The Department of Justice alleged that RealPage pooled competing landlords' nonpublic rents to generate recommendations, a "give to get" exchange of secret data. RealPage entered a proposed settlement in November 2025, and Greystar, the largest landlord in the country, settled with California for 7 million dollars in August. Earlier, in Gibson v. Cendyn Group, the Ninth Circuit held that nonbinding price recommendations are not a restraint of trade when they do not fold in competitors' confidential information.
The pattern across all of it is one distinction.
| Pricing input | Legal status in 2026 | Data source | Example |
|---|---|---|---|
| The individual shopper | Banned or label-required in a growing list of states | The buyer's own personal data | Raising a baby thermometer's price for a profiled new parent |
| Shared nonpublic competitor data | High antitrust risk; restricted and litigated | Rivals' secret prices pooled in a shared tool | A give-to-get rent algorithm fed by competing landlords |
| Public market signals | Protected, sometimes with a disclosure | Your own data and publicly posted prices | Matching a rival's listed price after demand shifts |
Read top to bottom, the safe zone is the bottom row.
Why This Raises the Value of Price Intelligence
Put the legal map next to the business reality and the move is obvious.
The personal-data method is getting expensive. It draws bans, civil penalties, attorney general sweeps, and a disclosure label that tells shoppers their own data set the price. The market-data method stays open. So the question for a pricing team is not whether to use data. It is which data.
Clean competitive price intelligence sits squarely in the protected zone. Here is why it travels well across every law on the books:
- It reads public prices, not private people. Tracking a competitor's listed price uses no shopper's browsing history or income, so the surveillance pricing bans do not reach it.
- It is your own tool, not a shared one. A single-firm system that monitors public listings does not pool rivals' nonpublic data, which is the conduct the antitrust rules and RealPage target.
- It supports market-based moves, the kind lawmakers protect. Adjusting to demand, inventory, and posted competitor prices is dynamic pricing, written into the New York bill as permitted.
There is a customer-trust dividend too. In a December 2025 Talker Research survey, 62 percent of shoppers said they were concerned about browsing history or location shaping their price. Nearly half, 48 percent, said they would favor a retailer that let them opt out of data-based pricing. Instacart ran a pricing test that reportedly produced up to 23 percent variation between customers for the same item. The backlash was fast, and the New York Attorney General sent a letter. Pricing from the market does not carry that risk. Pricing from the person does.
A note on the law itself. These statutes are new, they differ by state, and definitions are still moving. None of this is legal advice. Any brand building or buying a pricing tool should have counsel confirm how each rule applies to its data and its markets. Brands can see how we approach this in our pricing and competitive intelligence solutions.
The takeaway is not subtle. The risk was never using pricing data. The risk is using the wrong kind. Build your prices from the market, and the crackdown is not aimed at you. It just cleared your lane.