A Garbage Company, 17 Robocalls, and Why “Informational” Isn’t a TCPA Safe Harbor
On August 26, 2026, a federal court in the Middle District of Florida let a TCPA case go forward against a company most people would never picture in a robocall suit. Not a telemarketer. Not a lead-gen shop. A garbage collection company.
The case is Hartman v. Republic Services of Florida, LP, No. 5:25-cv-00790-PRL, 2026 WL 2500899 (M.D. Fla. Aug. 26, 2026). It's worth two minutes if you run any automated voice program, because it takes apart the single most common assumption operators make about the TCPA : that "informational" calls are safe.
They aren't.
What happened
Richard Hartman, proceeding pro se, alleged that Republic Services called his cell phone with a prerecorded voice about his garbage pickup. He says he never gave consent to be called with a prerecorded voice. He then mailed a cease-and-desist that read, in part: "Stop calling me. Do not call me at any number, for any reason."
Republic called at least 17 more times. Forty-five in all.
Hartman brought a single count under § 227(b)(1)(A)(iii). Republic moved to dismiss on four grounds. The court denied all four.
Four defenses, all rejected at the pleadings stage
On standing, Republic argued that informational service calls aren't the kind of harm the TCPA addresses. The court disagreed, citing Drazen v. Pinto, 74 F.4th 1336 (11th Cir. 2023) (en banc): an unwanted call is an intrusion on seclusion, and the analysis turns on the kind of harm, not the degree.
On consent, Republic argued that Hartman gave his number and therefore consented. If Hartman did give prior express written consent, then Republic didn't violate the TCPA But, that argument runs into the complaint itself. Hartman alleged he never provided the number and sent an unequivocal stop letter, and on a motion to dismiss the court takes those allegations as true.
On the prerecorded voice element, Republic argued the allegations were too thin. The court held that pleading the number of calls, their regularity, no human on the line, and a prerecorded message was enough to state a claim.
On revocation, Republic argued the letter was too vague and that an implicit revocation doesn't count. The court found "Stop calling me. Do not call me at any number, for any reason" clear enough to survive.
To be precise about posture: nothing here is decided. This is a denial of a motion to dismiss, not a verdict. But it means these theories were strong enough to put a garbage company into a year of TCPA litigation over service calls.
Why this matters for prerecorded and AI voice
Read § 227(b)(1)(A)(iii) closely. It bars an artificial or prerecorded voice call to a cell phone without prior express consent. It does not say "marketing calls." It says calls.
That's the gap most compliance programs live in. Appointment reminders, delivery windows, renewal notices, payment confirmations — operators treat these as exempt because they aren't selling anything. Under § 227(b), any prerecorded or artificial voice call to a cell still needs consent. Republic wasn't selling a thing and still lost its motion.
Now swap in an AI voice agent. "Artificial or prerecorded voice" is the exact statutory language (drafted in 1991, long before anyone built a conversational agent), and it describes precisely what an AI calling platform produces. The defendant in the next wave of these cases isn't a robocaller with a dialer. It's a company running a helpful-sounding automated voice without the right consent on file.
The revocation point operators miss
Consent is not a one-time checkbox. It can be revoked, and once it is, every call after is a fresh violation. Hartman's revocation wasn't a form or a portal submission. It was a plain "stop calling me," and the court still treated it as a plausible revocation under DNC rules.
That means a valid revocation can arrive in any words, through any channel — a voicemail, a reply text, a sentence to a call center, a letter — and if your systems don't catch it and push it to every platform that can dial, the calls that follow are the ones a court will treat as knowing. TCPA damages run $500 per call, up to $1,500 for willful. Seventeen calls after a written stop request is the kind of fact that reads as willful.
What to do this week
Two concrete moves. First, inventory every automated voice that touches a consumer phone, prerecorded and AI, and for each one confirm you can produce the consent to receive a prerecorded or artificial voice call — not just the phone number. Second, pressure-test your revocation path: send your own stop request through the messiest channel a real customer would use, then time how long until the calls actually stop across every system you run. If it's more than one call, you have Republic's exposure.
If you're running prerecorded or AI voice and you can't show, in five minutes, both the consent behind each call and the path a revocation takes through your stack, that gap is the exposure. We run flat-fee reviews that map consent and revocation across an automated-voice program to your actual systems. Reach out — happy to walk through your setup before a plaintiff's lawyer walks through it for you.
