The Supreme Court’s Narrow Definition

The Supreme Court settled the biggest question in TCPA litigation with a single unanimous decision. In Facebook, Inc. v. Duguid, decided April 1, 2021, the Court held that equipment only qualifies as an autodialer if it can store or produce telephone numbers using a random or sequential number generator. The full opinion is short, and the practical takeaway is shorter: dialing from a list does not make you an autodialer.

The ruling rejected the broader reading that lower courts had split over for years. Under that expansive view, the Second, Sixth, and Ninth Circuits would have swept in virtually every modern cell phone. The Court applied the series-qualifier canon and found that "using a random or sequential number generator" modifies both "store" and "produce," not just "produce."

The definition matters because the penalties are steep. The TCPA's private right of action allows recovery of $500 to $1,500 per call when a business uses an autodialer without consent. Volume alone proves nothing: the FCC clarified in its June 25, 2020 Declaratory Ruling that calling a large number of people is not probative of autodialer status.

What the ruling does not do is erase every obligation. The separate prohibition on artificial or prerecorded voices survives untouched, and consent remains a defense whether or not a system meets the autodialer definition. For anyone running structured outbound campaigns, the compliance posture is straightforward:

• Dial from pre-established, permissioned lists, never from generated numbers.
• Keep consent records on file even when the autodialer definition is not triggered.
• Treat AI-generated voices as artificial voices requiring prior express consent.
• Avoid random or sequential number generation anywhere in the dialing process.

Post-Duguid litigation has mostly held the line. In 2024, the Second, Third, and Ninth Circuits rejected arguments that using random number generation merely to sequence calls from a stored list creates ATDS liability. District courts still split on whether "capacity" means present ability or mere potential, and one court held that a randomly generated index number is not enough: the generated number must itself be a telephone number.

That is why My AI Call Center builds every campaign around approved, permissioned, or reviewed lists and checks list source and consent records before launch. Bought lists without clear permission records get flagged, and usually declined. The narrow definition rewards that discipline, and it punishes anyone still guessing.

Why Volume and List-Based Dialing Matter

If you're running thousands of calls a month, you might assume that volume alone puts you in autodialer territory. It doesn't. The rules that actually matter are about where your numbers come from and what kind of voice is on the call.

The FCC settled the volume question in a Declaratory Ruling on June 25, 2020, stating that "the fact that calling equipment is used to make calls or send texts to a large volume of telephone numbers is not probative of whether the equipment is an autodialer under the TCPA," according to the FCC's clarification. Scale is not the trigger. The trigger is whether your equipment stores or produces telephone numbers using a random or sequential number generator.

That distinction comes straight from the Supreme Court. In Facebook, Inc. v. Duguid, decided unanimously on April 1, 2021, the Court held that equipment which merely stores and dials numbers from a pre-established list does not qualify as an autodialer. The Court rejected the broader reading that would have swept in "virtually all modern cell phones," warning that such an interpretation "would take a chainsaw to these nuanced problems when Congress meant to use a scalpel."

For businesses running structured campaigns against known contacts, this matters enormously. Dialing from a curated, permissioned list is fundamentally different from generating numbers at random. But there are still details worth watching:

• Some courts have found that using a random generator to decide the order of calls from a preproduced list can still implicate autodialer liability.
• Others disagree. In *Tehrani v. Joie de Vivre Hospitality*, the court ruled that generating a random ID number to sequence a list is not enough: "the number generated must be a telephone number."
• Courts remain split on whether "capacity" means a system's current functions or its potential ability to generate numbers.
• Consent remains a defense regardless of autodialer status, so documented permission protects you either way.

Here's the part that catches people off guard: winning the autodialer question does not end your TCPA exposure. The prohibition on artificial or prerecorded voices is a separate legal risk that Duguid left completely untouched. If your calls use an AI-generated voice, that restriction applies on its own terms, and prior express consent is required.

This is why list discipline and voice compliance have to be handled together. At My AI Call Center, every campaign runs against approved, permissioned, or reviewed lists, with list source and consent records checked before launch, and AI-generated voices are treated as artificial voices requiring prior express consent. The statutory stakes are real: TCPA liability runs $500 to $1,500 per call. Getting both pieces right, the list and the voice, is what keeps a high-volume campaign on solid ground.

Consent and the Remaining Legal Risks

Even when your dialing technology clears the autodialer bar, you're not in the clear. The TCPA still gives plaintiffs other paths to liability, and consent remains the single most reliable defense.

The "capacity vs. use" debate continues to divide district courts. Some judges ask whether the equipment's current functions actually rely on a random or sequential number generator. Others ask whether the system merely has the capacity to do so, regardless of whether that capability is ever exercised. According to post-Duguid legal analysis, this split creates real uncertainty for businesses operating across multiple jurisdictions. One court may dismiss a case because the system wasn't actively generating random numbers. Another may allow the claim to proceed if the platform theoretically could be configured to do so.

The preproduced-list question adds another layer of confusion. Courts generally agree that dialing from a curated contact list does not make your system an autodialer. But when a platform uses a random or sequential generator to determine the order in which it pulls numbers from that list, results vary. In Tehrani v. Joie de Vivre Hospitality LLC, the court ruled that generating a random or sequential index number to assign dialing order does not trigger ATDS classification, because the statute requires the generated number to be a telephone number, not just an identifier. Other courts have reached the opposite conclusion. The safest path is to avoid random or sequential number generation anywhere in your dialing workflow.

Plaintiffs have noticed the narrowing. Post-Duguid, they are shifting toward TCPA provisions that don't require ATDS use at all, including fax rules and do-not-call registry violations. Courts have been more receptive to expansive readings in these areas, which means the autodialer definition is no longer the only battleground.

This is why prior express consent remains the primary defense. The FCC confirmed in its June 2020 Declaratory Ruling that even when an entity uses an autodialer, it can avoid liability by obtaining the recipient's prior express consent. With TCPA damages running $500 to $1,500 per call, that defense is worth far more than any technical argument about dialing equipment.

For organizations running structured outbound campaigns, the practical playbook looks like this:

• Dial only from approved, permissioned, or reviewed lists where consent records exist and can be produced if challenged.
• Confirm your dialing platform does not use random or sequential number generation for telephone numbers, and document that fact.
• Treat AI-generated and prerecorded voices as triggering separate consent requirements, since the *Duguid* ruling left that prohibition untouched.
• Honor DNC requests immediately and carry opt-outs across all current and future campaigns.

My AI Call Center applies this framework before any campaign launches: list source and consent records are checked, bought lists without clear permission are flagged or declined, and the team tells you plainly if a list won't support the campaign before you spend anything. The autodialer question matters, but consent is what actually keeps you safe.