Skip to main content

Is AI Cold Calling Legal? TCPA Rules Explained

AI cold calling is legal in the US when it follows TCPA consent and disclosure rules. Here is what the FCC's 2024 ruling changed, which state laws go further and how to keep your outreach compliant.

Is Ai Cold Calling LegalAi Cold Calling TcpaAi Voice Calls LegalTcpa Ai Calling RulesAi Sales Calls Compliance
Deepak Singh
Deepak Singh 8 min read
Share this post
Is AI Cold Calling Legal? TCPA Rules Explained

AI cold calling is legal in the United States. The catch is the same one that has governed automated calls for years: you need the right consent, you have to identify yourself, and you have to honor opt-outs. The February 2024 FCC ruling did not outlaw AI calling. It confirmed that an AI-generated voice counts as "artificial" under the Telephone Consumer Protection Act (TCPA), so it lives under the rules that already applied to robocalls.

That distinction got lost in the headlines, which left a lot of sales teams convinced they were suddenly breaking the law. They were not, at least not for putting AI on the phone. They would be breaking it by calling people who never agreed to hear from them, with or without AI in the mix.

This guide walks through what the ruling actually changed, when AI cold calling is legal, where it crosses the line, and how to keep your outreach compliant while you use AI calling tools to build pipeline. Get it wrong and the math is brutal: TCPA damages run $500 to $1,500 per call, with no cap on the total.

Key takeaways

  • AI cold calling is legal with consent. The FCC classified AI voices as "artificial" under the TCPA, so marketing calls need prior express written consent.
  • The penalties are real. TCPA violations cost $500 to $1,500 per call, and there is no ceiling on a class action.
  • State law often goes further. Florida's FTSA sets shorter calling hours, caps call frequency and hands consumers a private right of action.
  • Compliance is a system, not a hope. Screening every number before you dial (safe for AI, human-only, or do-not-call) keeps a campaign clean from the first ring.

What the FCC's 2024 ruling actually changed

On February 8, 2024, the FCC adopted a unanimous Declaratory Ruling that treats AI-generated voices as "artificial" under the TCPA. No new statute, no new agency, just a clarification that voice-cloning and synthetic-voice calls sit under the rules Congress already wrote.

The useful way to read it: AI calling is not illegal; calling people without their consent is, and an AI voice does not change that.

Per the FCC's official announcement, the ruling was aimed at scammers cloning voices to impersonate relatives, celebrities and politicians, and it gave state attorneys general a clear hook to chase those cases. For a legitimate B2B team, it is almost reassuring. If your robocall practices were already TCPA-compliant, your AI calling is too. The standard did not move. The technology just caught up to it.

When AI cold calling is legal

Three things have to be true before an AI-voiced sales call is on solid ground.

You have prior express written consent. For marketing calls placed with an artificial voice, the recipient has to have agreed, in writing, to receive automated calls from you. Electronic signatures and checked consent boxes count. A purchased list does not, because whatever consent sits on that list was given to someone else. Picture a SaaS team that buys 5,000 phone numbers and points an AI Call Agent at them the next morning. Every one of those calls is a potential $500 to $1,500 violation, because not one of those people consented to hear from that company. Existing-customer relationships and certain non-marketing calls can clear a lower bar (prior express consent, not necessarily written), but cold B2B sales almost always lands in the stricter written-consent category.

The call identifies who is behind it. Every AI call has to state the name of the business placing it and give a number the recipient can call back. Your AI Call Agent should say who it is and which company it represents in the opening line, not three sentences in.

People can opt out, easily. Consumers can revoke consent through any reasonable means, and businesses have to act on it fast. Under the TCPA opt-out rules, a revocation has to be processed as soon as practicable and no later than 10 business days, and you cannot force people through a single approved channel to do it. If someone says stop, by voice, text or email, the calling stops.

When AI cold calling crosses the line

AI calling turns illegal the moment one of the basics is missing.

The most common failure is the obvious one: no consent. Dialing an artificial voice at someone who never opted in violates the TCPA, and buying a list does not transfer consent to you. After that it is a short set of bright lines. Numbers on the National Do Not Call Registry are off-limits for telemarketing, AI or human, so lists have to be scrubbed against it before a campaign goes out. Cell phones carry extra protection and generally need consent before any automated call. Federal rules bar telemarketing before 8 a.m. or after 9 p.m. in the recipient's local time, and some states pull that window in tighter. And the Truth in Caller ID Act makes it illegal to spoof misleading caller ID with intent to defraud, so your outbound has to show real, accurate information.

None of these are AI-specific. They are the rules of the road for any automated outreach. AI just makes it easier to break all of them faster.

State law usually raises the bar

Federal TCPA is the floor. A wave of state "mini-TCPA" laws has built a second, often stricter layer on top of it, and they apply based on where your prospect sits, not where you do. If you call into multiple states, the strictest one on your list effectively sets your rules, so a state-by-state view of mini-TCPA laws is worth keeping close.

Florida is the one that reshaped the industry. The Florida Telephone Solicitation Act (FTSA) shortens calling hours to 8 a.m. through 8 p.m., caps you at three calls within a 24-hour period to the same person on the same subject, and, critically, gives individuals a private right of action with statutory damages of $500 per violation, trebled to $1,500 when the violation is willful or knowing. That last part is why the state became a magnet for class actions. California adds its own layer: a state do-not-call list, two-party consent for recording calls and CCPA obligations on the consumer data you hold.

The trend is not slowing. States including Oklahoma and Washington have passed their own telemarketing statutes in recent years, and several require solicitors to register before they dial. The practical takeaway is simple: clearing the federal boxes is not enough. You have to clear compliance in every state where your prospects actually live.

Can you cold call with AI in California?

Yes. AI cold calls are legal in California under the same federal floor, prior express written consent for an AI voice, honest caller ID and an opt-out honored on the spot, plus the state layer above: California's do-not-call list, all-party consent before a call is recorded or monitored, and CCPA obligations on the personal data your campaign holds. In practice that means the agent discloses itself and any recording in the first breath of the call, and your suppression list carries both the federal and state DNC entries. Consent, disclosure and clean lists get you through.

What non-compliance really costs

TCPA penalties add up fast:

Violation TypePenalty Per Call
Standard TCPA violation$500
Willful TCPA violation$1,500
State mini-TCPA violationUp to $1,500

The per-call number is only the opening line. Enforcement has teeth now: the FCC has signed robocall-investigation partnerships with attorneys general in 49 states, plus the District of Columbia and Guam, pooling resources to track down illegal callers. Class actions are the bigger threat, because there is no cap on statutory damages. A single campaign to 10,000 non-compliant numbers is $5 million in exposure at the floor and $15 million if the violations are ruled willful. And the quietest cost never shows up in a settlement: a prospect who gets an unwanted, non-compliant call almost never becomes a customer. The trust you burn outlasts any deal you might have closed. For the full picture, see the financial and reputational fallout of getting this wrong.

How to keep AI calling compliant

The rules are knowable, which means compliance is a system you build once, not a judgment call you make on every dial.

It starts before anyone picks up the phone. Every number should be classified up front into one of three buckets: safe for an AI call (consent on file, not on a do-not-call list, no other restriction), human-only (something needs a person's judgment, like an unclear consent record), or do-not-call (listed, no consent, or otherwise blocked). AvairAI's one-click TCPA compliance system runs that classification on every number before a campaign launches, so the risky ones never get dialed.

From there, the rest is record-keeping and timing. Keep proof of how and when each consent was obtained, what the person agreed to, and when any opt-out was processed; those records are your defense if a complaint ever lands. Scrub lists against the National Do Not Call Registry, the relevant state lists and your own internal do-not-call list before every send. Show accurate caller ID, your real company name and a working callback number. And let the system handle calling windows: it should detect each recipient's time zone and schedule around both the federal 8 a.m. to 9 p.m. rule and tighter state cutoffs like Florida's 8 p.m.

The compliant way to use the phone is Pair Selling

Here is the part most "AI cold caller" pitches skip. US law effectively limits automated AI calling to warm or opted-in contacts, so aiming an artificial voice at a cold list is both legally risky and the wrong tool for the job. The channel that still works on a cold prospect is a human one.

That is the logic behind Pair Selling, and it happens to be the compliant logic too. AvairAI builds and runs the campaign: it finds the right accounts, writes the personalized outreach, sends the emails automatically and screens every phone number for TCPA compliance before anyone dials. Your reps walk into ready-to-run call and LinkedIn tasks, each with the contact and a personalized script in hand. Where automated calling fits, it stays where the law keeps it, on warm and opted-in contacts, always disclosed as AI. This hybrid of AI and human calling is what compliant phone outreach actually looks like.

The division of labor is the point. AvairAI surfaces interested leads and clears the compliance work; your reps have the conversations that book and close. You are not choosing between moving fast and staying legal. The same system does both.

The bottom line on AI cold calling

So, is AI cold calling legal? Yes, when it is done right. The February 2024 FCC ruling did not rewrite the rules. It confirmed that an AI voice is "artificial" under the TCPA and owes the same consent and disclosure standard as any other automated call.

For a B2B team, that comes down to a handful of habits: get genuine written consent before you place AI calls, classify every number before you dial, learn the state laws where your prospects live (Florida's FTSA above all), document everything and use tooling that builds compliance in instead of bolting it on. For a deeper operational walkthrough, the TCPA compliance playbook for sales leaders covers the full workflow.

The teams winning with AI on the phone are not the ones testing how close to the line they can get. They are the ones treating compliance as the foundation of trust, because a prospect who knows you respect the rules is a prospect more likely to pick up. Ready to run outreach that is compliant from the first call? Start your free campaign and let the compliance work run itself.


← Back to all articles
Deepak Singh

About Deepak Singh

CEO & Co-founder, AvairAI

Deepak Singh is the CEO and co-founder of AvairAI, pioneering "Pair Selling" — AI agents that run B2B prospecting while salespeople focus on closing. He brings 25+ years as a founder and technology leader: he co-founded enterprise-software company Adeptia in 2000 and served as CTO and President through 2025, building a data-integration/iPaaS platform for mission-critical connectivity and earning a US patent for his B2B-connectivity invention. Earlier he led product at 3Com (scaling its cable-modem business to $40M), Netscape, and AMD. He holds an MS in Engineering from Stanford, an MBA from Northwestern’s Kellogg School, and a BS in EECS from UC Berkeley. An InfoWorld-quoted voice on AI agent architecture, he writes widely on building and scaling companies, AI sales implementation, and RevOps.

More from Deepak Singh →

See what AvairAI builds from your website

Never sell alone.

14-day free trial · no credit card · see it in ~3 minutes

Prefer to browse first? Grab a free outreach template Start for free