Three places a franchise system can now be held to account for its AI, and only one of them is the technology

The debate about AI in franchising is usually framed as a capability question. Which agent, which vendor, which use case. The record of 2026 suggests the harder question is a liability question, and that it lands on paper rather than on code. Three surfaces have now produced live exposure.

Surface one: pricing

According to PYMNTS on 8 October 2026, summarising Reuters reporting from 5 and 8 October, a federal antitrust class action was brought against McDonald's alleging that the company pressures franchisees to use its AI-powered algorithmic pricing, and that this drives up consumer prices. McDonald's position, as reported, is that each franchisee sets its own prices and that AI does not set the price of a Big Mac or any other menu item.

The same report notes something structurally new. The suit may be the first to bring such claims against franchisees of the same brand rather than against competing companies. That matters because the standard antitrust intuition assumes rivals. A franchise system is not a set of rivals in that sense, and the law has not finished deciding what a shared pricing recommendation means inside one brand.

PYMNTS also records that the case law is unsettled in both directions. One recent appeals ruling said pricing software enabled collusion that would have been harder in the past. Another held that using the same price-recommendation software was not by itself enough for an antitrust claim. The practical reading for a franchisor is that a pricing recommendation sits in an area where the outcome is not predictable yet, and that the wording of how strongly it is recommended may decide the case.

Surface two: the vendor contract

The second surface is the commercial paper under the technology. According to Restaurant Dive on 30 October 2026, the US Federal Trade Commission and eight states sued Toast over hidden fees, auto-renewing subscriptions averaging USD 900 a month, and allegedly false claims about free hardware and round-the-clock support. The claimed violations include the FTC Act and the 2010 Restore Online Shoppers Confidence Act, and the suit says at least tens of thousands of consumers complained.

That case is not about AI. But it is about the contract through which AI features are sold, and in a franchise system that contract usually has two parties on the buyer side in substance and one on paper. The franchisor specifies the stack. The franchisee signs it and pays monthly. If a specified vendor turns out to have terms a regulator objects to, the franchisee holds the exposure and the franchisor holds the specification decision.

Surface three: what AI says to a prospect

The third surface is the one franchisors talk about most and disclose least. Franchise Times, on 29 May 2026, reported that a Franchise Insights survey found nearly 90% of franchisors used AI for recruitment content in 2025 or planned to start in 2026, with nearly 80% using or planning to use it for recruitment ad images and text, and nearly 90% expecting to use at least one AI tool by the end of 2026. Only 28.2% used AI to contact or qualify leads in 2025, with 26.9% planning to start in 2026.

The same report carries the performance numbers that explain the adoption. Batteries Plus chief development officer Joe Malmuth described a 400% increase in qualified appointments on the AI side within 30 days of a test, and the company shortened the time from informational form to signed franchise agreement to about 75 days from 120. Senior Helpers, with operators in more than 420 US territories, spends USD 50,000 to USD 100,000 monthly on digital marketing and is piloting AI to pre-screen the more than 100,000 digital inquiries it receives a year.

Here is the exposure. Every claim an AI system makes to a prospective franchisee is a claim the franchisor made. A faster funnel is a larger volume of statements, generated at a speed no human reviewed. Franchise disclosure regimes care about financial performance representations regardless of which system produced the sentence.

The one question that covers all three

Across pricing, vendor terms and recruitment, the common defence is the same: a written record of who decided, what was recommended, how strongly, and what the prospect was told. That is a documentation discipline, not a technology investment. A system that cannot produce that record is exposed on all three surfaces at once, however good its tools are.

What we do not know

The McDonald's case has not been decided, and PYMNTS reports no dollar figures or store counts. We do not know whether a court will treat a franchisor pricing recommendation as pressure.

The Toast complaint names no franchise system and no franchisee, so we cannot say how much of the alleged cost fell on franchised units rather than independents. Toast has said it will defend itself, and its argument that the 2010 statute does not apply to negotiated business-to-business agreements has not been tested.

On recruitment, the Franchise Insights figures are self-reported survey intentions from a May 2026 report, not audited adoption. And no source in this set reports a single enforcement action or private claim arising from AI-generated franchise recruitment material. The exposure described here is structural, not yet demonstrated.

This article is information, not legal or investment advice.

Get next week's brief by email. Free.