AI Phone Spam Is at a 6-Year High — and AI Calls Are Legally Robocalls

Your phone is ringing more than it did last year — and there's a good chance the voice on the other end isn't human. AI has made robocalls cheaper, more convincing, and harder to ignore, and regulators have responded with a clear rule: if it sounds like a call and acts like a call, it's a call under the law.

The numbers back up the frustration. According to YouMail data cited by U.S. PIRG, scam and telemarketing robocalls hit roughly 2.56 billion per month through September 2025 — a 20% jump over 2024 and the highest level in six years. About 57% of all robocalls this year are scam or telemarketing calls, and 31% of American adults report getting at least one scam call every single day.

What makes AI-driven spam especially dangerous is believability. As consumer advocates warn, AI deepfakes can now fool people into thinking they're talking to a close relative or lifelong friend in real time. The average scam-call victim lost $3,690 in the first half of 2025, according to FTC data.

Here's the part most businesses miss: the FCC has definitively settled the legal question. In its 2024 Declaratory Ruling, the agency confirmed that AI-generated voice calls fall under the Telephone Consumer Protection Act's restrictions on "artificial or prerecorded voice." The statute, as the FCC put it, "does not allow for any carve out of technologies that purport to provide the equivalent of a live agent."

In plain terms, that means:

• Every AI outbound call needs prior express consent, no matter how human the voice sounds.
• Marketing AI calls generally require written consent; informational calls can rest on oral consent, per TCPA compliance analysis.
• An established business relationship does not cover AI calls — a live agent can dial a past customer under EBR; an AI agent cannot without separate consent.

Legal experts at Manatt, Phelps & Phillips agree: if a call uses an artificial or prerecorded voice, the caller must obtain the called party's prior express consent unless an exemption applies. The stakes are real — TCPA violations run $500 to $1,500 each, and class-action filings are up 95% year over year.

This is why list discipline matters more than ever. At My AI Call Center, every campaign runs only against approved, permissioned, or reviewed lists, with list source and consent records checked before a single call goes out. In an environment where the FCC treats AI voice calls as illegal without consent, that pre-launch review isn't just good practice — it's the difference between a useful call and a violation.

Why the Wrong List Gets You Sued: Consent, EBR, and the $500–$1,500 Per-Call Risk

Why the Wrong List Gets You Sued: Consent, EBR, and the $500–$1,500 Per-Call Risk

A single misstep in consent can turn a helpful AI call into a costly legal liability. Many businesses assume that an existing customer relationship grants permission for automated outreach — but with AI-generated voices, that assumption is dangerously flawed.

Under the TCPA, AI-powered calls are classified as artificial or prerecorded voice calls, triggering strict consent requirements that vary by call type and jurisdiction. For informational or transactional calls — such as appointment reminders or payment notifications — prior express consent is required, which can be obtained orally. However, for marketing or advertising AI calls, prior express written consent is mandatory in 47 states, with only Texas, Louisiana, and Mississippi allowing oral consent for marketing purposes under Fifth Circuit precedent. This tiered consent architecture is not optional; it is the foundation of lawful AI calling.

One of the most expensive misunderstandings in AI outbound calling is the belief that an Established Business Relationship (EBR) overrides consent requirements. As legal experts emphasize, EBR does not cover AI calls — a live agent may contact a past customer under EBR, but an AI agent cannot dial the same person without separate, specific consent. “The voice is what the law cares about,” meaning the artificial nature of the voice eliminates traditional exemptions that apply to human-led outreach.

The financial risk of getting this wrong is severe. Each TCPA violation carries a penalty of $500–$1,500 per call, with penalties rising to $1,500 if the violation is deemed willful. Class-action exposure is growing rapidly, with TCPA filings up 95% year over year and settlements frequently reaching the $5M–$20M range. Notable cases include Gen Digital’s $9.95M settlement and QuoteWizard’s $19M resolution, underscoring how quickly liability accumulates when consent is mishandled.

Liability also extends through the vendor chain. The Lamb v. Mortgage One Funding precedent establishes that businesses using third-party AI calling services remain responsible for compliance — assuming the vendor owns the risk is a critical error. This means list quality, consent documentation, and opt-out honoring must be verified at every stage, not delegated.

• AI-generated voice calls require prior express consent under TCPA, with marketing calls needing written consent in 47 states
• Established Business Relationship does not exempt AI calls from consent requirements
• TCPA violations cost $500–$1,500 per call, with class-action filings up 95% YoY

For organizations using managed AI calling services like My AI Call Center, compliance begins long before the first dial — it starts with rigorous list and consent review, transparent documentation, and a clear understanding that permission, not presumption, governs every outbound AI interaction.

The Safeguards That Keep AI Calls Off the Spam Pile

The difference between a useful AI call and a spam complaint usually comes down to a handful of operational details. With roughly 2.56 billion scam and telemarketing robocalls per month hitting American phones through September 2025, carriers and consumers are quick to flag anything that feels evasive. The safeguards below are what keep legitimate AI campaigns off that pile.

AI disclosure on every call. The FCC has confirmed that AI-generated voices count as artificial voices under the TCPA, no matter how human they sound. Several states go further: Texas SB 140 requires disclosure within 30 seconds of call start, while Florida requires written consent that explicitly references AI use, with California, Colorado, Illinois, and Utah each adding their own variants. A clear opening line — "This is an AI assistant calling on behalf of [Company]" — is fast becoming standard practice, and recipients should always be able to ask whether the call is AI-assisted, request a human, or opt out.

Keyword opt-outs, honored immediately. When a recipient says "STOP" or "REVOKE," the call should end and the number should be suppressed on the spot — not after a manual review next week. Logging every opt-out, and carrying it into the client's permanent do-not-call record, closes the gap that turns one unwanted call into a formal complaint. With TCPA penalties running $500 to $1,500 per violation, a sloppy opt-out log is an expensive liability.

DNC requests across all campaigns. Federal rules require do-not-call and consent revocation requests to be honored within 10 business days, but well-run operations treat that as a ceiling, not a target. A suppression request on one campaign should apply to every campaign that follows. This is why My AI Call Center runs list and consent reviews before launch, checks permission records on every list, and declines bought lists that lack clear documentation — the opt-out and DNC logs ship with every campaign's completion report.

Quiet hours and calling windows. State-specific quiet hours, day restrictions, and registration rules vary, and calls must respect the recipient's time zone, not the caller's. Structured campaigns lock calling windows before launch — after-hours leads get queued and called first thing the next business day rather than dialing outside approved hours.

The core safeguards, in short:

• Disclose AI status on every call, within state-specific time limits
• Honor STOP and REVOKE keywords immediately, with a full audit log
• Respect DNC requests across all campaigns within 10 business days — ideally sooner
• Call only inside approved, state-compliant windows

None of this is glamorous. It is disciplined operational hygiene, and as compliance analysts note, an established business relationship does not exempt AI calls — the voice itself is what the law regulates. Campaigns built on approved, permissioned lists with these safeguards in place are the ones that get answered, not blocked.

How My AI Call Center Stops Spam Before It Starts

Most AI spam doesn't start with a bad script — it starts with a bad list. That's why every campaign we run begins with list and consent review before a single call goes out.

The stakes are real. TCPA penalties run $500–$1,500 per violation, and class-action filings are up 95% year over year, with settlements landing in the $5M–$20M range. The FCC has confirmed that AI-generated voices count as "artificial or prerecorded voice" under the TCPA, so prior express consent is required no matter how human the voice sounds.

Here's how the safeguards work in practice, step by step.

List and consent review comes first. Before any campaign launches, we review the list source, consent records, and calling windows. Consent tiers matter: marketing AI calls generally require prior express written consent in 47 states, while informational calls can rest on oral consent. A list without clear permission records can't support a compliant campaign — so we check before anything else does.

Bought lists get flagged — and usually declined. Purchased lists without documented permission records are flagged, and in most cases we decline them outright. We tell you plainly if the list won't support the campaign, before you spend anything. That's cheaper for everyone: a declined list costs nothing, while a non-compliant campaign can cost six figures in a single settlement.

Script and escalation approval before launch. The script, the AI disclosure, the opt-out handling, and the escalation path all go to you for sign-off. Nothing launches until you approve. Every call includes AI disclosure, and recipients can ask for a human or opt out by keyword.

Opt-outs and DNC logs ship with every report. Every outcome report includes disposition codes, per-call notes, and complete opt-out and DNC logs. Opt-outs are logged and honored immediately, and DNC requests are respected across all campaigns and carried into client DNC records — well within the 10-business-day window required by TCPA amendments.

The result is a simple rule that governs everything:

• Only approved, permissioned, or reviewed lists — never indiscriminate cold calling
• Consent records verified before launch, not after a complaint
• Plain feedback when a list won't work, before you spend anything
• Opt-out and DNC logs delivered with every outcome report

That's how My AI Call Center stops spam before it starts — by refusing to launch the calls that become it.

Your Pre-Launch Checklist: Run Compliant AI Campaigns from 9¢ Per Minute

The difference between a campaign that gets results and one that gets you sued is almost always what happened before the first call was dialed. With TCPA penalties running $500 to $1,500 per violation and class-action filings up 95% year over year, your pre-launch checklist matters more than your script.

Here is the checklist we run through before any campaign goes live — and the one you should demand from any AI calling partner.

1. Verify list source and consent records. Every contact on the list must trace back to a documented permission record. Bought lists without clear permission records get flagged and, in most cases, declined. As legal analysis from Manatt confirms, AI calls are treated as artificial voice under the TCPA, so the caller must have prior express consent unless an exemption applies. Retain those records for at least the four-year statute of limitations.

2. Match consent tier to call type. This is where most organizations slip. Informational calls generally require only prior express consent, which can be oral — but marketing calls require prior express written consent in 47 states, per the TCPA compliance playbook. And remember: an established business relationship does not cover AI calls the way it covers live agents dialing past customers.

3. Confirm state disclosure rules. Federal AI disclosure mandates are still pending, but states are not waiting. Texas requires disclosure within 30 seconds under SB 140, and California, Florida, Colorado, Illinois, and Utah each have their own variants — Florida even requires written consent that explicitly references AI use.

4. Set approved calling windows. Quiet hours and day restrictions vary by state and time zone, so windows must be set per jurisdiction before launch, not adjusted after complaints arrive.

5. Document opt-out handling. Opt-outs must be logged and honored immediately, with keyword handling for terms like STOP and REVOKE, and DNC requests carried across all campaigns into your permanent DNC records. The 2025 TCPA amendments require honoring revocations within 10 business days — immediate is simply better.

None of this is optional anymore. With robocalls at a six-year high and regulators shutting down nearly 1,400 non-compliant phone companies in 2025, the enforcement environment rewards organizations that can prove their discipline on paper.

At My AI Call Center, this checklist is built into every campaign review — nothing launches until you approve the script, disclosure, opt-out handling, and escalation path. Campaigns run against approved, permissioned, or reviewed lists only, from 9¢ per connected minute, with the full number known before launch.

Ready to run a compliant campaign? Book a free campaign review via Plan My Campaign — we will tell you plainly whether your list and consent records will support the goal before you spend anything.