A Michigan mortgage lender is facing a federal class-action lawsuit over AI-generated calls that pitched refinancing to a consumer whose number was on the National Do Not Call Registry (MPA, 2026). **Using AI to call a customer who hasn't reached out first is legally the same as any other telemarketing robocall — it needs its own written consent, or the business is exposed to $500 to $1,500 in statutory damages per call.** The calculus is different for a business that only uses AI to answer calls customers place to it.
What Happened in the Mortgage One Funding AI Voice Lawsuit?
Filed February 24, 2026 in the U.S. District Court for the Eastern District of Michigan, the class action accuses Mortgage One Funding, a Clawson, Michigan-based lender, of running a wide-scale telemarketing campaign that used an artificial voice to cold-call consumers' cell phones pitching cash-out refinancing, without prior express written consent (Scotsman Guide, 2026). The complaint alleges the call opened by falsely claiming to be returning a previous call that never happened, and that the recipient's number had been on the National Do Not Call Registry since 2023; the plaintiff says the AI voice was identifiable by "awkward pauses and odd vocal inflection" before the call transferred to a live sales representative (MPA, 2026). If the class is certified and the company is found liable, exposure runs $500 per violation, rising to $1,500 if a court finds the violations were willful and knowing — across a proposed nationwide class of anyone who received a similar call from the lender or its vendors and lead generators (Scotsman Guide, 2026).
Is It Legal to Use AI to Call Customers for Marketing or Sales?
Yes, but only with the same prior express written consent required for any prerecorded or autodialed marketing call under the TCPA — the FCC's February 2024 ruling confirmed an AI-generated voice counts as an "artificial voice" under that law, so the consent requirement doesn't drop just because the voice sounds more natural than an old-style robocall (Henson Legal, 2026). What trips businesses up most is assuming an existing customer relationship covers it. Henson Legal's compliance guidance is explicit: an Established Business Relationship exempts a business from Do-Not-Call list rules for a manual, human-dialed call, but it does not exempt an AI-generated voice call from TCPA consent requirements — the AI voice itself is what triggers the obligation, regardless of how well the business already knows the customer (Henson Legal, 2026).
| Lower-risk (existing consent likely covers it) | Higher-risk (needs its own written consent) | |
|---|---|---|
| Example call | Appointment reminder to a patient or client who booked and gave a phone number for that purpose | Cold call to a purchased list or old lead pitching a new product |
| Consent status | Tied to the specific transaction the customer initiated | No prior relationship or AI-specific consent on file |
| Do Not Call Registry exposure | Lower — transactional/informational, not telemarketing, when scoped correctly | High — cold marketing pitches are exactly what the registry targets |
| What Mortgage One Funding is accused of | Not this | This — a cold pitch, a DNC-registered number, a false "returning your call" opening (Scotsman Guide, 2026) |
Does This Risk Apply to an AI Receptionist That Only Answers Calls?
No — and that distinction gets lost in most coverage of this lawsuit. TCPA consent rules govern calls a business places to a consumer; the law has nothing to say about a customer calling a business's own number and an AI system answering it (Henson Legal, 2026). A plumbing company's AI phone system that picks up an inbound call, books the job, and texts a confirmation isn't making an outbound telemarketing call — it never dialed anyone. The Mortgage One Funding case, and the compliance framework built around it, applies specifically to businesses dialing out: appointment-reminder campaigns, review-request calls, cold sales outreach, win-back calls to a stale lead list. A business using AI only to answer its own phones can read this story as background, not a warning.
What Should a Business Do Before Using AI to Call Customers?
- Get consent in writing, specific to AI or artificial-voice calls — a generic "you agree to be contacted" checkbox written before AI voice existed doesn't cover it (Henson Legal, 2026).
- Scrub every outbound calling list against the National Do Not Call Registry on a recurring basis, not just once when the list is built (Henson Legal, 2026).
- Keep the AI's opening honest — a call that claims to be "returning" a call that never happened is itself part of what's being litigated in the Mortgage One Funding case (MPA, 2026).
- Know that a vendor's mistake becomes the business's liability — TCPA exposure follows calls made by a company's lead generators and vendors, not just calls it dials directly (Henson Legal, 2026).
- Keep dated proof of consent on file — if a demand letter arrives, a business needs to show exactly when and how a customer agreed to be called this way.
A two-location dental or med-spa group running between Kalispell and Missoula, or a property manager working both sides of the Flathead Valley, is exactly the kind of multi-location Northwest business likely to want AI for appointment reminders and review-request calls once it's already using AI to answer its phones — and exactly the kind of business this case is a warning to, not the plumbing outfit that only uses AI to pick up. The fix isn't avoiding outbound AI calling; it's building the consent and list-hygiene layer before turning it on, not after a demand letter shows up.