The Autodialer Definition That Actually Controls Your Campaigns

For years, "autodialer" was the most expensive undefined word in telemarketing. Courts couldn't agree on what it meant, and businesses dialed anyway, hoping they were on the right side of it. The Supreme Court settled the question in April 2021, and the answer is narrower than most people expect.

In Facebook, Inc. v. Duguid, the Court ruled unanimously that an automatic telephone dialing system must have the capacity to store or produce telephone numbers using a random or sequential number generator, and to dial those numbers. The phrase "using a random or sequential number generator" applies to both "store" and "produce." Equipment that merely stores and dials numbers from an existing list does not qualify.

That reasoning resolved a six-circuit split. The Second, Sixth, and Ninth Circuits had read the definition broadly; the Third, Seventh, and Eleventh read it narrowly. The Court sided with the narrow view, rejecting an interpretation that would have swept in everyday smartphones. Justice Sotomayor wrote that the broad reading would "take a chainsaw to these nuanced problems when Congress meant to use a scalpel."

So where does list-based calling land? Outside the definition. Predictive dialers that dial from pre-existing lists may not qualify as ATDS under the ruling, because stored lists don't involve random or sequential number generation. That matters for any campaign built on a real contact list rather than generated numbers, which is why the equipment question is only the first gate we check when reviewing a campaign's list source and consent records before launch.

Here's the part that trips people up: winning the ATDS question does not win the TCPA. The statute works through several independent prongs, and each one can create liability on its own:

• Prerecorded or artificial voice calls, which remain violations regardless of how the number was dialed
• Calls to numbers on the National DNC Registry, which holds over 250 million phone numbers
• Calls made without the required consent, with damages of $500 to $1,500 per call and no aggregate cap
• Calls placed outside permitted windows, before 8:00 AM or after 9:00 PM in the called party's local time

The artificial-voice prong deserves special attention for anyone using AI callers. On February 8, 2024, the FCC confirmed that AI-generated voices, including voice cloning, fall within "artificial or prerecorded voice" under the TCPA, and it explicitly rejected any carve-out for technologies claiming to be "the equivalent of a live agent." Your dialer might sit safely outside the autodialer definition while your voice sits squarely inside the artificial-voice rules.

The practical takeaway is simple. The Duguid test is equipment-based and narrow, so list-dialing clears it. But consent, DNC status, calling windows, and voice technology each carry their own obligations, and several states layer on TCPA-like laws of their own. Clearing the narrowest test in the statute is a starting point, not a compliance plan.

Why AI Voice Calls Face a Separate Consent Track

Here's the part most AI calling operators miss: even after the Supreme Court narrowed the autodialer definition, your AI campaigns still face full TCPA liability through a completely separate door. On February 8, 2024, the FCC issued a Declaratory Ruling confirming that AI-generated voices, including voice cloning, fall under the TCPA's definition of "artificial or prerecorded voice" (Wilson Sonsini). The reasoning is blunt: a person is not speaking, so the voice is artificial.

The FCC also closed the loophole everyone hoped for. It explicitly rejected any carve-out for technologies claiming to be the "equivalent of a live agent," stating the TCPA does not allow for it (FCC Declaratory Ruling, FCC 24-17). How human your AI sounds is irrelevant. What matters legally is how the voice was produced.

That classification puts every AI voice campaign on the artificial-voice compliance track, which means:

• Prior express consent is required, and AI-generated robocalls to wireless numbers require prior express written consent.
• The Established Business Relationship exemption does not cover AI calls. EBR exempts you from DNC Registry restrictions, but the artificial voice itself still triggers the consent obligation.
• State AI disclosure rules already apply in Texas (within the first 30 seconds), California, Florida, Colorado, Illinois, and Utah.
• Penalties run $500 to $1,500 per call with no aggregate cap, and TCPA class action filings are up 95% year over year.

The EBR point deserves repeating because it's counterintuitive. A live SDR can dial a 16-month-old customer on the DNC list under EBR. Your AI agent cannot call that same person without separate consent. The voice is what the law cares about.

This is why the list review step matters so much. At My AI Call Center, list source and consent records get checked before any campaign launches, and bought lists without clear permission records are flagged or declined outright. "Warm cold lists" and co-registration leads have no legal standing for AI outbound without re-consent, no matter how recent the last purchase was. When we scope a campaign, we treat it as an artificial-voice campaign first and a calling campaign second, because that's the order the law applies.

Disclosure is the other piece. The FCC's ruling arrived alongside a proposed federal rule (still pending as of April 2026) that would require in-call AI disclosure and consent language specifically referencing AI use. Several states aren't waiting for that rule; Florida already requires written consent that references AI. Building AI disclosure and opt-out handling into every script now, rather than retrofitting it later, keeps campaigns ahead of both the FCC and the state patchwork.

The practical takeaway: don't assume your dialing setup avoids the autodialer definition and therefore avoids the TCPA. The artificial-voice prong applies to every AI call you place, independently, and it carries the same $500 to $1,500 per-call exposure.

The Compliance Gates That Apply No Matter How You Dial

You might assume that clearing the autodialer definition solves your compliance puzzle. It doesn't. The TCPA operates through multiple independent gates that trigger liability regardless of your dialing equipment.

Even if your system dials from a static list without a random generator, you still face strict consent requirements. The FCC confirmed in 2024 that AI-generated voices are "artificial" under the law, triggering specific consent mandates detailed in the Declaratory Ruling. This means your voice technology creates its own legal obligations.

Consent records must be granular and auditable. The FCC’s one-to-one consent rule prohibits blanket forms covering multiple sellers. Each record needs a timestamp and specific permission details. A single missing data point can expose you to statutory damages ranging from $500 to $1,500 per call as outlined in current compliance analyses.

DNC scrubbing is a non-negotiable operational rhythm. You must check your lists against the National Do Not Call Registry at least every 31 days. The Registry contains over 250 million numbers, so a single missed scrub creates significant exposure.

Calling windows are hard stops, not suggestions. Federal rules prohibit calls before 8:00 AM or after 9:00 PM in the recipient's local time zone. State laws often tighten these windows further, requiring you to apply the most restrictive rule for each specific number dialed.

Opt-out handling demands immediate action. As of April 2025, you must honor opt-out requests within 10 business days. Ignoring this window creates a separate violation, distinct from the initial consent failure.

State mini-TCPA laws add another layer of complexity. Washington and Connecticut have broader definitions of automated dialing than federal law. Texas requires AI disclosure within the first 30 seconds of a call. You must map these state-specific requirements to every number in your campaign.

My AI Call Center treats these gates as hard parameters. We review list sources and consent records before launch. If a list lacks clear permission records, we flag it and often decline the campaign. This approach ensures that every call runs within approved windows and honors state-specific quiet hours.

The compliance burden is real and fragmented. You cannot rely on a single federal standard. You must manage consent, DNC status, timing, and state rules simultaneously.

• Consent records must be granular, timestamped, and auditable per individual call.
• DNC scrubbing must occur at least every 31 days to remain compliant.
• Calling windows vary by state, with some requiring calls to end by 8:00 PM.
• Opt-out requests must be honored within 10 business days.
• State laws may require specific AI disclosure within the first 30 seconds.

Ignoring these independent gates creates liability even if your dialer is technically compliant. The risk is cumulative. Each violation adds to your potential statutory damages.

The legal landscape is shifting, but the core obligations remain stable. You need systems that enforce these rules automatically. Manual checks are too slow and prone to error.

Your campaign setup should include state-by-state configuration for windows and disclosures. This prevents accidental violations in stricter jurisdictions. It also protects your reputation with customers who expect respectful communication.

Compliance is not a one-time audit. It is an ongoing operational discipline. You must monitor opt-outs, update DNC lists, and verify consent records continuously.

The cost of non-compliance far exceeds the cost of proper setup. A single campaign with systemic consent deficiencies can result in millions in damages. Protecting your business requires treating these gates as critical infrastructure.

Ready to run compliant campaigns? Plan your next campaign with a focus on approved, permissioned lists. We handle the compliance gates so you can focus on the conversation.

Building a Campaign That Holds Up to the Narrow Definition

Here's the thing most teams get backwards: they spend weeks asking whether their dialer meets the ATDS definition, and almost no time on the records that actually decide lawsuits. After Facebook v. Duguid, equipment that merely stores and dials numbers from a list doesn't qualify as an autodialer at all (as Duane Morris notes). But that narrow test doesn't save you when the claim is about consent, artificial voice, or the DNC Registry.

Your real defense is list discipline. A campaign of 100,000 calls with a systemic consent deficiency represents $50 to $150 million in potential statutory damages, since violations run $500 to $1,500 per call with no aggregate cap (per compliance analysis). That's why the practical work starts before the first call, not after the first complaint.

Review every list at the source level. Ask where each number came from and what permission sits behind it. "Warm" lists and co-registration leads have no legal standing for AI outbound without re-consent, and the established business relationship exemption doesn't cover AI calls because the artificial voice itself triggers the consent obligation (as one TCPA playbook puts it). A bought list without clear permission records should be flagged or declined outright. My AI Call Center applies exactly this screen before any campaign launches, because a list that can't support the campaign costs more than it earns.

Document consent to the one-to-one standard. The FCC's 2024 rule means a single blanket consent form can no longer cover multiple sellers; records must be granular, timestamped, and auditable at the individual call level (per recent compliance analysis). For each contact, you should be able to answer three questions on demand:

• When did this person consent, and to which seller?
• What did the consent language actually say, and did it reference AI or prerecorded voice where state law requires it?
• Can you produce the record for any single call in the campaign within minutes?

Configure state rules per number, not per campaign. Federal windows prohibit calls before 8:00 AM or after 9:00 PM local time, but Florida, Oklahoma, and Washington all cut off earlier, and Oklahoma caps you at three calls per number per 24 hours regardless of consent (per federal rules research). Washington and Connecticut also define autodialing more broadly than federal law, so apply the most restrictive applicable rule based on the called party's state. Scrub against the National DNC Registry at least every 31 days, and honor opt-outs within 10 business days, the tightened standard as of April 2025.

Keep records for seven years. The TCPA statute of limitations is four years, but defense counsel recommend seven years of retention (per the same playbook). And remember that liability runs up the vendor chain: proposed classes have reached everyone who received artificial-voice calls from a company's vendors, lead generators, or agents. Your name is on the call even when someone else sourced the list.