Why Voice AI Falls Under the TCPA (And Why It Can't Be Dodged)

Many businesses assume that if their AI voice sounds human enough, it escapes TCPA scrutiny—but the FCC has definitively ruled otherwise. AI-generated voices are explicitly classified as "artificial or prerecorded voice" under 47 U.S.C. § 227(b), regardless of how lifelike they sound, closing any perceived loophole around "live agent equivalent" claims. This classification triggers the same consent requirements as traditional robocalls, meaning prior express consent is required for informational calls and prior express written consent for marketing calls in most jurisdictions.

The financial exposure is severe and growing rapidly. Statutory TCPA damages range from $500 to $1,500 per call with no aggregate cap, so even a modest 10,000-call campaign could face $5 million to $15 million in statutory exposure before actual damages are considered. Class-action litigation is surging, with filings up 95% year-over-year and aggregate verdicts exceeding $925 million. High-profile settlements like QuoteWizard’s $19 million payout underscore the real cost of non-compliance, especially when businesses mistakenly rely on established business relationships—which do not exempt AI calls from consent requirements under the TCPA.

For companies using managed services like My AI Call Center, this means compliance cannot be outsourced or assumed. Entities commissioning AI calls remain liable for vendor actions, as confirmed in litigation such as Lamb v. Mortgage One Funding. Critical obligations include delivering AI disclosure within the first 30 seconds of the call (required in TX, CA, FL, CO, IL, and UT), honoring opt-outs within 2 seconds of the initial message, and maintaining auditable consent records. State-level rules further complicate compliance, with calling windows varying from 8 AM–8 PM in Florida and Oklahoma to 9 AM–9 PM Monday–Saturday and noon–9 PM Sunday in Texas. Ignoring these nuances isn’t just risky—it’s financially untenable in today’s litigation-heavy environment.

• AI disclosure must occur within the first 30 seconds of the call to comply with state laws in Texas, California, Florida, Colorado, Illinois, and Utah.
• Opt-out requests must be honored within 2 seconds of the initial message to avoid TCPA violations.
• Prior express written consent is required for marketing AI calls in most states, though Texas, Louisiana, and Mississippi now allow oral consent post-Bradford ruling.

The Consent Rules That Actually Apply to AI Calls

Many businesses mistakenly assume that using AI for calls bypasses traditional consent rules, but the FCC has made it clear: AI-generated voices are treated as artificial or prerecorded voice under the TCPA, triggering the same consent requirements as any robocall. This means consent isn't optional—it's the foundation of legal outbound AI calling.

For informational or transactional calls—such as appointment reminders, payment notifications, or service confirmations—prior express consent (PEC) is sufficient. This can be obtained orally, in writing, or electronically, as long as it's clear the consumer agreed to receive such calls. However, for marketing or telemarketing calls—including lead qualification, upsell offers, or promotional surveys—prior express written consent (PEWC) is required in most states. PEWC must include a signature, whether electronic or written, and clearly authorize the specific type of calls being made.

There is an important exception in Texas, Louisiana, and Mississippi. Following the Bradford v. Sovereign Pest Control ruling in February 2026, oral consent now suffices for marketing calls in these three states. This means a simple, recorded verbal agreement—such as "Yes, I agree to receive promotional calls from your company"—can meet the PEWC standard where federal law would otherwise require written documentation. This creates a critical compliance distinction based on geography that voice AI deployments must account for.

One of the most dangerous misconceptions in AI calling is that an established business relationship (EBR) eliminates the need for consent. The research is unequivocal: EBR only provides relief from calling numbers on the National Do Not Call Registry—it does not waive TCPA consent requirements for AI-generated voice calls. Relying on EBR as a substitute for proper consent has led to numerous class-action lawsuits, with settlements in 2025–2026 ranging from $5M to $20M and statutory damages of $500 to $1,500 per call. Even a modest 10,000-call campaign could face $5M–$15M in statutory exposure before actual damages are considered.

• Informational calls require prior express consent (PEC), which can be oral, written, or electronic.
• Marketing calls require prior express written consent (PEWC) in most states.
• Texas, Louisiana, and Mississippi allow oral consent for marketing calls post-Bradford ruling.
• Established business relationships (EBR) do not exempt AI calls from TCPA consent requirements.
• Statutory TCPA damages range from $500 to $1,500 per call with no aggregate cap.

For organizations using managed AI calling services like My AI Call Center, this means consent verification isn't just a legal formality—it's a core part of list hygiene and campaign safety. Every number dialed must be backed by verifiable consent matching the call type and jurisdiction, or the entire campaign faces significant legal and financial risk. Getting consent right isn't about checking a box—it's about building trust and ensuring every call is both useful and lawful.

Disclosures, Opt-Outs, and State-by-State Rules You Must Honor

Operational compliance is where voice AI deployments live or die, with specific timing requirements that leave little room for error. The FCC and multiple states now mandate that AI disclosure must occur within the first 30 seconds of a call—a rule actively enforced in Texas, California, Florida, Colorado, Illinois, and Utah. This isn't merely a suggestion; it's a legal trigger that, if missed, can invalidate consent and open the door to statutory damages ranging from $500 to $1,500 per call. For a service like My AI Call Center, which manages campaigns for clinics, franchises, and membership organizations, embedding this disclosure into the opening seconds of every script is non-negotiable, especially when calling permissioned lists where trust is already established.

Equally critical are the mechanics of opt-out processing and revocation honors. Systems must recognize and act on keywords like "STOP" or "REVOKE" within two seconds of the initial message, ensuring the call terminates or transfers immediately without further engagement. Once an opt-out is received, TCPA requires that revocation be honored within 10 business days—a timeline reinforced by FCC provisions effective April 2025. Failure here represents one of the three dominant litigation patterns identified by legal experts: consent gaps, late disclosure, and failed STOP processing. These aren't theoretical risks; they've fueled settlements exceeding $925 million in recent years, with individual cases like QuoteWizard reaching $19 million due to vendor chain liability failures.

Calling windows add another layer of state-specific complexity that must be baked into campaign scheduling. While the federal baseline allows calls from 8:00 AM to 9:00 PM in the recipient's local time, several states impose stricter limits. Florida, Oklahoma, and Oregon restrict calls to 8 AM–8 PM, while Connecticut allows 9 AM–8 PM. Texas presents a split schedule: 9 AM–9 PM Monday through Saturday, and noon–9 PM on Sundays. These variations mean a national campaign can't rely on a single time window—it requires dynamic adjustment based on the recipient's area code and number portability data. For organizations running multi-touch campaigns across healthcare reminders or renewal notices, getting this wrong isn't just a compliance misstep; it's a direct path to class-action exposure where even 10,000 non-compliant calls could yield $5M–$15M in statutory damages before actual harm is proven.

How We Build TCPA-Ready Voice AI Campaigns

Knowing the rules is one thing; running campaigns that survive four years of litigation exposure is another. That's why compliance isn't a checkbox at the end of our process — it's the first gate every campaign passes through.

Before any call goes out, we review the list source and consent records. Bought lists without clear permission records get flagged, and in most cases declined. We tell you plainly if the list won't support the campaign, before you spend anything. This matters because legal analysis shows the three dominant violation patterns in active litigation are consent gaps, missing or late AI disclosure, and revocation failures — all problems that start with the list, not the technology.

Every call discloses that it's AI-assisted. States including Texas, California, Florida, Colorado, Illinois, and Utah require AI disclosure within the first 30 seconds, and compliance guidance suggests a single clear sentence satisfies most jurisdictions at once. Recipients can ask if the call is AI, request a human, or opt out.

Opt-outs are honored immediately. When a caller says STOP or REVOKE, the request is logged and carried into your DNC records across all campaigns. Federal rules now require revocation to be honored within 10 business days, but waiting that long invites risk — immediate is the only acceptable standard.

Calling windows are state-aware, not just federal-aware. The federal baseline is 8 AM–9 PM local time, but state rules narrow that considerably: Florida, Oklahoma, and Oregon cap calls at 8 PM, Connecticut at 8 PM starting at 9 AM, and Texas restrict Sunday calls to noon–9 PM. Campaigns adjust for the recipient's state, not the caller's convenience.

Every campaign closes with full documentation:

• Dispositioned contact list with per-call outcomes
• Opt-out and DNC logs tied to every campaign
• Completion and coverage reports
• Escalation paths and script approvals on record

Here's the part most vendors won't say out loud: if you commission the calls, you share the liability. Courts have confirmed that companies hiring AI-calling vendors remain responsible for vendor actions. With statutory damages of $500 to $1,500 per call and no aggregate cap, a 10,000-call campaign gone wrong means $5M–$15M in exposure before actual damages. Defense counsel recommend keeping records for seven years — which is why documentation isn't paperwork, it's your defense file.

That shared liability is exactly why list discipline is non-negotiable at My AI Call Center. A campaign that can't show where its list came from and what consent backs each number isn't a campaign we'll run — no matter how good the script is.

Your Pre-Launch TCPA Checklist for Voice AI

Before your first AI call goes out, the cheapest compliance fix is the one you make in a planning document — not the one you make in response to a demand letter. With statutory damages running $500 to $1,500 per call and no aggregate cap, a 10,000-call campaign carries $5M–$15M in potential exposure, per legal analysis. A pre-launch checklist turns that risk into a manageable process.

1. Verify consent records match the call type. Informational calls like appointment reminders need prior express consent (PEC), but marketing calls require prior express written consent (PEWC) in most states. One wrinkle: after the February 2026 Bradford v. Sovereign Pest Control ruling, Texas, Louisiana, and Mississippi accept oral consent for marketing calls. Your consent records must document which type you captured — and for what purpose.

2. Confirm state calling windows and disclosure rules. The federal baseline is 8 AM–9 PM in the recipient's local time, but at least five states are stricter: Florida, Oklahoma, and Oregon cap calls at 8 PM, Connecticut starts at 9 AM, and Texas limits Sunday calls to noon–9 PM, according to state-by-state compliance guidance. Texas, California, Florida, Colorado, Illinois, and Utah also require AI disclosure within the first 30 seconds.

3. Test opt-out handling before launch. The three dominant litigation patterns are consent gaps on transfers, late or missing AI disclosure, and revocation failures — the AI simply doesn't recognize "STOP." Opt-out must activate within 2 seconds of the initial message, and revocation must be honored within 10 business days under FCC rules effective April 2025, per compliance counsel. Run test calls that try to opt out every way a real person would.

4. Plan your record retention. TCPA claims carry a four-year statute of limitations, and defense counsel recommend retaining records for seven years, per industry guidance. Your retention plan should cover:

• Consent capture records, including timestamp, source, and disclosed purpose
• Call logs with dispositions, disclosures delivered, and opt-out requests
• Script versions and escalation paths, dated and approved
• DNC and revocation records carried across all campaigns

5. Get legal guidance for regulated industries. Healthcare, financial services, and employment screening stack additional rules on top of TCPA — Colorado's pending ADMT framework will reach voice AI used for consequential decisions in lending, insurance, and employment by June 2026. If your list touches a regulated area, involve counsel before launch, not after.

The starting point is a campaign review: define the one clear goal, then check list source, consent records, and calling windows against it. My AI Call Center runs this review free before any campaign, and flags lists that won't support the goal before you spend anything. Keep in mind that requirements vary by location, industry, contact type, and consent status — a checklist is a starting point, not a substitute for legal advice specific to your situation.