Complying with the NY Algorithmic Pricing Disclosure Act

Industry: Retail & Sales | Audience: General Counsel / Chief Pricing Officer

Direct Answer

The New York Algorithmic Pricing Disclosure Act (effective 2025) requires retailers to disclose when artificial intelligence or algorithmic systems materially influence pricing shown to consumers. Violators face civil penalties of up to $500 per day per violation, with no cap for continuing offenses. If your pricing engine uses AI, machine learning, dynamic segmentation, or automated demand forecasting to set or adjust prices, you must disclose that fact to consumers at or before the point of sale — or restructure your pricing workflow to remove algorithmic decision-making from customer-facing outputs. General Counsel and CPOs have until Q1 2025 to audit, remediate, and operationalize compliance before the law takes effect and before copycat legislation in 12+ states creates a patchwork of conflicting obligations.

Executive Reality

Most retailers have no idea how deep algorithmic pricing runs in their stack. The pricing model that adjusts your e-commerce tags based on competitor scraping? Algorithmic. The A/B test that shows different prices to different ZIP codes? Algorithmic. The demand-forecasting module that recommends promotional markdowns? Algorithmic. The surge-pricing logic on your delivery app? Algorithmic. The ERP rule that applies "dynamic elasticity adjustments" to seasonal SKUs? Algorithmic.

The NY law defines "algorithmic pricing" broadly: any system that uses "computational processes, including artificial intelligence, machine learning, or data analytics, to set, recommend, or substantially influence the price of a good or service offered to a consumer." This is not limited to third-party SaaS tools. An Excel macro that pulls competitor prices and feeds your ERP qualifies if it "substantially influences" the final price.

The disclosure requirement is triggered at the point where the consumer "commits to the transaction" — which means product listing pages, checkout flows, and in-app confirmations. A generic privacy-policy footnote does not satisfy the law. The disclosure must be "clear and conspicuous" — defined as unavoidable, plain-language notice that specifically identifies the price as algorithmically set or influenced.

Cost of Inaction

The $500/day penalty is per violation — meaning per transaction. A mid-market retailer processing 10,000 algorithmically priced transactions daily faces $5M per day in exposure. The law includes a private right of action, so expect enforcement mills to batch thousands of violations into class complaints.

Root Cause

Algorithmic pricing grew organically without legal architecture. Every function — revenue management, e-commerce, demand forecasting, competitive intelligence — bought or built its own pricing tool. No central inventory of pricing algorithms exists. Legal reviewed contracts, not logic. Procurement approved SaaS spend, not data flows. The result: pricing decisions are algorithmically influenced at dozens of touchpoints, and no single executive can trace the full chain from data input to price output.

The NY law is not an anomaly. It is the first enforcement wave of a global trend: algorithmic transparency mandates. The EU AI Act classifies certain pricing systems as "high-risk." The DMA requires gatekeepers to explain ranking and pricing mechanisms. Brazil, Canada, and the UK have consultations open. Treating NY as a one-off compliance checkbox is a strategic error.

Framework: The Algorithmic Pricing Compliance Playbook

Phase 1: Discovery (Weeks 1–4)

Objective: Build a complete inventory of every system, model, rule engine, and workflow that influences customer-facing prices.

1. System Census. Interview owners of every system with pricing touchpoints: e-commerce platform, POS, ERP, revenue management, competitive intelligence, promotion engine, loyalty program, marketplace feed, delivery app, mobile app.
2. Data Flow Mapping. Trace each system's inputs (competitor data, demand signals, inventory levels, customer segments) to pricing outputs.
3. Algorithmic Classification. Tag each system as: (A) fully algorithmic — AI/ML sets the price; (B) partially algorithmic — recommends price, human approves; (C) rules-based — static logic, no learning; (D) human-only — no computational influence.
4. Materiality Assessment. Determine which systems "substantially influence" final customer-facing prices. This is the legal threshold for disclosure.

Phase 2: Remediation Design (Weeks 5–8)

For each materially influential algorithmic system, choose one of three paths:

Phase 3: Disclosure Engineering (Weeks 9–12)

Design disclosures that satisfy legal requirements without destroying conversion:

• Placement: At or before commitment — PDP, cart, checkout confirmation, or in-app price breakdown.
• Wording: Plain language — "This price was set using automated technology that analyzes market demand and competitor pricing" — not legal boilerplate.
• Granularity: Per-item for mixed carts where some items are algorithmically priced and others are not.
• Format: Inline text, not hover-over or buried in terms of service.
• Accessibility: Screen-reader compatible; meets WCAG 2.1 AA.

Phase 4: Multi-State & EU Expansion (Weeks 13–20)

Map NY requirements against pending legislation in other jurisdictions. Build a compliance matrix that tracks each state's definition of "algorithmic," disclosure trigger, penalty structure, and private right of action. Align with EU DMA obligations if you operate in or sell into Europe.

Minimum Viable Action (MVA)

This week:

5. Schedule a 90-minute pricing-system census with representatives from e-commerce, revenue management, legal, and IT. Ask one question: "What systems influence the final price a customer sees?" Map every answer.
6. Flag your highest-transaction-volume pricing algorithm. If you had to add a disclosure to one system today, which touches the most customers? That is your Priority 1 remediation target.
7. Draft a placeholder disclosure for that system. Run it past legal and test it with 5 customers. If conversion drops more than 3%, pivot to the "De-algorithmize" or "Hybrid" path.

Risk Register

What Not To Do

• Do not rely on your privacy policy. The law requires transaction-level, point-of-sale disclosure. A policy page footer is insufficient.
• Do not wait for final implementing regulations. The statutory text is clear enough to act on. "We'll wait for regulatory guidance" is how you get penalized on day one.
• Do not assume SaaS vendors handle this. Your vendor's compliance posture does not transfer liability. You are the retailer; you face the penalty.
• Do not hide disclosure in hover text or sub-menus. "Clear and conspicuous" has a legal standard. Courts will test whether a reasonable consumer saw it before committing.
• Do not de-algorithmize everything. Full reversion to static pricing sacrifices competitive positioning and margin. Make targeted, materiality-driven decisions.
• Do not treat this as a purely legal exercise. The CPO, CRO, and CTO must co-own implementation. Legal can define the threshold; operations must engineer the fix.

Scale-or-Stop

Continue IF: You have completed the system census, identified materially influential algorithms, and selected a remediation path (Disclose, De-algorithmize, or Hybrid) for your top 3 highest-transaction-volume systems.

Stop and reassess IF: You cannot identify which systems influence pricing, or if your organization lacks the technical capability to modify price-display logic. In that case, engage external counsel and a systems integrator immediately — do not self-certify compliance you cannot verify.

Scale IF: Your NY remediation is operational, tested, and documented. Use the same playbook for pending legislation in other states and EU DMA alignment. Treat this as a global algorithmic transparency infrastructure, not a one-jurisdiction patch.

FAQs

Q: Does the law apply to B2B pricing? A: The NY Act applies to "consumers" — defined as individuals acting for personal, family, or household purposes. B2B transactions are outside the scope unless the buyer is a sole proprietor purchasing for personal use. However, pending legislation in other jurisdictions may expand this. Monitor broadly.

Q: What if our algorithm only recommends a price, and a human approves it? A: The law covers systems that "recommend or substantially influence" pricing. If human approval is rubber-stamp (approval rate >90%, review time <30 seconds), regulators will likely classify this as algorithmic. Document the genuine human judgment applied.

Q: Does third-party marketplace pricing on our platform trigger our disclosure obligation? A: If you operate the marketplace and set or influence fees/commissions that affect the final consumer price, likely yes. If you are purely a passive platform with no pricing influence, likely no — but marketplace operators should seek specific legal counsel.

Q: Can we satisfy disclosure with a generic "prices may vary" statement? A: No. The law requires specific identification of algorithmic influence. Generic disclaimers do not meet the "clear and conspicuous" standard.

Q: What is the penalty for a single transaction to a single consumer? A: The law specifies "per violation" — interpreted as per instance of non-compliant pricing. A single consumer transaction is one violation, subject to $500/day if the non-compliance continues without remediation.

Q: Does the law apply to promotional pricing and discounts? A: If the promotional price or discount percentage is set or influenced by an algorithm, yes. Static "20% off all items this weekend" set by a human planner does not trigger disclosure.

Q: Should we lobby for amendments? A: Through trade associations, yes — but do not make lobbying your primary strategy. Comply with the law as written while advocating for clarity. Banking on legislative relief is not a risk management strategy.

Final Recommendation

The NY Algorithmic Pricing Disclosure Act is the opening salvo in a multi-year, multi-jurisdiction campaign to force transparency into automated commercial decisions. Complying with NY is non-negotiable. But the strategic opportunity is broader: retailers that build algorithmic transparency as a consumer trust signal will outperform those that treat it as a compliance burden.

My recommendation: Complete your pricing-system census within 30 days. For each materially influential algorithm, choose Disclose, De-algorithmize, or Hybrid based on transaction volume, margin sensitivity, and brand risk. Engineer point-of-sale disclosures that are legally sufficient and conversion-optimized. Document every decision. Then scale the playbook across all jurisdictions where you operate.

The penalty for delay is not a fine. It is a $500-per-transaction-per-day litigation magnet that plaintiff firms will mine systematically. Move now.