紐約市議會推進禁止以個人資料設定個人化價格的法案
Same Ride, Different Price: New York City Moves to Ban Prices Set From Your Personal Data
紐約市議會推動 Intro 891-A,擬禁止企業利用個人資料和演算法設定或調整個人化價格;議會官員稱法案有足夠票數通過,但仍須市長簽署,並會在成為法律一年後生效。法案涵蓋裝置追蹤、瀏覽紀錄、生物特徵資料及購買紀錄,並保留一般促銷、優惠券和符合披露條件的忠誠獎賞等做法。違規罰款最高為每項 $1,000,小型企業首次違規可先獲通知並有 30 天改正。
Same ride. Same route. Same time of day. Different price, depending on whose phone is asking.
That isn't a conspiracy theory from a group chat. According to amNY, a New York City Council investigation used 10 rideshare accounts to run about 370 price checks on three routes. Some accounts kept getting automatic discounts the others didn't. At least one account never got a discount at all.
On November 12, 2026, the Council was set to vote on Intro 891-A, a bill that bans what it calls surveillance pricing. A Council official told amNY the bill had enough votes to pass. It still needs the mayor's signature, and it would take effect one year after it becomes law.
I've spent 13+ years building web products and helping founders launch them, so I want to look at this from the builder's side of the screen.
What the bill actually bans
The original Council announcement describes it as stopping businesses from using personal data to set individual prices, fees and discounts. That data includes device tracking, browsing history, biometric data and purchase history.
In the revised version reported by amNY, companies can't use personal data and algorithms to set or adjust individualized prices. They also can't collect, sell or share personal data to make that possible.
What stays allowed:
- Ordinary promotions and coupons
- Loyalty rewards, including tiered ones, if businesses disclose how they work
- Discounts for groups like seniors and veterans
- Some purchase-history discounts, with disclosure and limits on data use
- Regulated areas such as insurance and credit pricing
Enforcement goes to the city's Department of Consumer and Worker Protection. Fines go up to $1,000 per offense, and each affected customer or transaction can count separately. Small businesses get a notice and 30 days to fix a first violation. There's no private right to sue, which the department's own commissioner had asked for.
The bill also grew from two pages to ten after retail, hotel and delivery groups objected that the first draft would sweep up normal loyalty deals.
Why this is easier to build than you'd think
Here's the uncomfortable part for people like me. Building "this user sees a different price" is not an engineering moonshot.
A product already knows your device, your location, how often you open the app, and whether you abandoned a cart last week. Turning that into a price is a few rules or a small model sitting between the catalog and the checkout page. A/B testing tools already do something similar for button colors.
That's why I don't buy the argument that this is too complex to regulate. If it's simple enough to build in a sprint, it's simple enough to describe in a law.
The case against the ban (and it's not nothing)
I try to be fair here, because I've sat on the business side of pricing decisions.
Dynamic pricing isn't automatically evil. Cheaper off-peak tickets, last-minute hotel deals and surge pricing that pulls more drivers onto the road can all help customers. A first-time-buyer discount is technically a personalized price too.
The industry's worry was that a blunt ban would kill the good stuff along with the creepy stuff. Judging by the four-page jump to ten, the Council seems to have listened.
There's also a fair question about enforcement. A $1,000 fine per violation can add up fast, but proving that an algorithm used your data to charge you more is hard when you can only see your own screen.
Where I draw the line
My rule of thumb as a founder is simple: if you'd be embarrassed to show the customer why they got that price, don't ship it.
A discount someone can see, understand and ask about is a deal. A price set quietly from data the customer never knowingly handed over is a bet on how badly they need the thing. The first builds trust. The second works right up until a journalist or a city council runs 370 tests.
And trust is the cheapest marketing a small product will ever get. Big platforms can absorb a scandal. Most startups can't.
A quick check if you build products
If you run pricing, discounts or checkout logic, it's worth an hour this week:
- List every input that can change the final price a user sees.
- Mark which of those are personal data (location, device, history, behavior) and which are public rules (coupon codes, published tiers, time of day).
- For each personal input, ask whether you could explain it to the user in one sentence on the checkout page.
- If not, either drop it or make it a visible, opt-in discount.
Even if you're nowhere near New York, rules like this tend to spread. Europe is already debating its own fairness rules for digital products.
Your turn
I'd personally take one honest price for everyone over winning or losing an algorithm lottery I can't see.
But I'm curious where you'd draw it. Is a personalized discount fine as long as it's disclosed, or should the price be the same for everyone, full stop? Why?
來源:DEV Community · dev.to