AI Dynamic Pricing Discrimination Law 2026: Surveillance Pricing Legal Risk
AI pricing engines that set a different price for every shopper are drawing FTC attention and new state legislation. If your business uses algorithmic or "personalized" pricing, here's where the legal exposure actually sits.
What Changed: From Dynamic Pricing to Surveillance Pricing
Dynamic pricing — adjusting prices based on supply, demand, time of day, or broad market segments — has been legal and common for decades (airlines, ride-share, hotels). What's new is AI's ability to price at the individual level, using data about a specific shopper: their device, location, browsing history, past purchases, and inferred willingness to pay. Regulators now draw a line between segment-level dynamic pricing and individual-level "surveillance pricing," and the latter is where scrutiny is concentrated.
The FTC's study into surveillance pricing practices found that several companies maintain the technical capability to change prices in real time based on a specific consumer's data, even when they don't always use it. Regulators have signaled that the existence of this capability, not just active use, is part of what they're scrutinizing — which matters for any business building or buying an AI pricing engine with that capability built in.
Where AI Pricing Crosses Into Discrimination
The AI Pricing Compliance Checklist
- ☐List every data input your pricing algorithm uses — device, location, browsing history, purchase history, referral source, account tenure
- ☐Flag any input that correlates with a protected characteristic even indirectly (zip code, device tier, language settings)
- ☐Test pricing outputs across demographic groups for disparate impact, not just disparate treatment
- ☐Document the business justification for any input that could function as a proxy
- ☐Disclose when a price shown to a consumer has been personalized, per emerging state disclosure requirements
- ☐Avoid design patterns that make personalized pricing look like a standard list price
- ☐Keep records of what price was shown to which consumer and why, for audit and dispute purposes
- ☐Monitor California and New York surveillance pricing legislation and comparable bills in other states
- ☐Check whether your pricing data practices trigger CCPA obligations around automated decision-making and profiling
- ☐Coordinate pricing compliance review with privacy compliance review — they draw on the same underlying data flows
- ☐If you license a third-party AI pricing engine, ask the vendor directly what data inputs and targeting granularity it supports
- ☐Require contractual disclosure of any protected-class-correlated inputs the model uses or could use
- ☐Disable individual-level targeting capability by default; enable only with a documented compliance review
Why This Is Different From Hiring or Lending Discrimination
Algorithmic hiring and lending discrimination have established legal frameworks (Title VII, ECOA) and years of enforcement precedent. Algorithmic pricing discrimination is earlier-stage: there's no single federal pricing-discrimination statute equivalent to ECOA, so exposure comes from a patchwork — FTC Act Section 5 unfair-practices authority, state consumer protection and price-gouging statutes, emerging state surveillance-pricing bills, and disparate-impact theories borrowed from other discrimination law.
That patchwork makes it easy to underestimate the risk, because there's no single checklist analogous to an ECOA compliance program. Businesses deploying AI pricing at scale should treat this the way early-stage hiring-algorithm compliance was treated before NYC Local Law 144 and Colorado's AI Act arrived — get ahead of the audit requirement before a regulator or plaintiff's attorney forces the issue.
Frequently Asked Questions
Does surge pricing count as surveillance pricing?
No, if it's based on aggregate real-time supply and demand (e.g., ride-share surge pricing applied uniformly to everyone in a zone) rather than an individual consumer's personal data. Surveillance pricing specifically refers to individualized pricing based on a specific person's data, not broad demand-based adjustments applied to everyone equally.
We don't intentionally target any protected class in our pricing model. Are we safe?
Intent isn't the full test. Disparate impact claims focus on outcomes: if your model's inputs correlate closely enough with a protected characteristic that pricing outcomes differ systematically by that characteristic, you can face legal exposure even without intent. This is why input auditing and output testing across demographic groups both matter.
What's the difference between price discrimination and price discrimination law?
Economically, 'price discrimination' just means charging different customers different prices — a neutral, common, and often legal practice. Legally, 'price discrimination' becomes actionable when it violates specific statutes: discrimination against a protected class, state consumer protection law, antitrust law (like the Robinson-Patman Act for certain seller-to-seller pricing), or new surveillance-pricing disclosure rules. Most everyday dynamic pricing is legal; the risk is concentrated in the individual-targeting, proxy-data, and non-disclosure scenarios.
Related Guides
Audit the Inputs Before Regulators Do
The fastest way to reduce AI pricing discrimination risk is to audit what data feeds the model, not just what price it produces. Proxy variables are the recurring failure mode — pricing inputs that never touch a protected characteristic directly but correlate with one closely enough to produce the same effect.
Get the input audit and disclosure practices in place now, while state surveillance pricing law is still forming — retrofitting compliance after a state law or FTC action lands is far more expensive than building it in from the start.