Skip to content
tuvoz

Is it legal for an AI to call a business?

Yes, with rules: US law treats AI voices as artificial voices, and some states add rules on automated calls, recording and AI disclosure. General information, not legal advice, as of Oct 8, 2026.

· Published Oct 8, 2026 · 12 sources

Illustration: balanced scales with a purple phone handset on one pan and a small storefront on the other, in front of a pale shield.

Yes, in the US an AI can call a business for you, but the call is regulated. Federal rules treat an AI voice as an “artificial voice”, which means the call must say who is responsible for it, and some states add their own rules on automated calls, on recording, and on telling people they’re talking to an AI.

This page is general information, not legal advice. It reflects what we found as of Oct 8, 2026, and laws in this area are changing. If your situation matters (you’re calling for a business, at volume, or for sales), ask a lawyer.

The short answer, in five points

  1. AI voices count as artificial voices. In February 2024 the FCC ruled that AI-generated voices are “artificial” under the Telephone Consumer Protection Act (TCPA) (FCC 24-17).
  2. Every artificial-voice call must identify who’s responsible and give a number. At the start of the message the call must “state clearly the identity of the business, individual, or other entity that is responsible for initiating the call”, and during or after the message it must “state clearly the telephone number” of that business or person, which can’t be the number that placed the call (47 CFR 64.1200(b)).
  3. Consent rules depend on the line. Artificial-voice calls to mobile phones need the called party’s prior express consent, and so do calls to home lines unless an exemption applies (47 U.S.C. 227(b)(1)). A business landline called for an ordinary errand is outside those two consent rules. Toll-free numbers can fall under the mobile-phone rule, which also covers “any service for which the called party is charged for the call”.
  4. Some lines are off limits. The same section bars artificial-voice calls without consent to emergency lines, including 911 and the emergency lines of hospitals, poison control, and fire and police agencies.
  5. States add more. Some states regulate any call that uses a recorded or synthesized voice, about a dozen require everyone’s consent to record a call, and a few have AI disclosure laws. See AI call disclosure rules by state.

What makes a business call lower risk

  • The line is a business landline, not a mobile phone or someone’s home.
  • The purpose isn’t sales. Booking a table, asking about stock, rescheduling an appointment. Telemarketing brings in stricter written-consent rules and the Do Not Call registry.
  • The call says it’s an AI, who it’s for, and who’s placing it, at the start.
  • It asks before recording, which matters in states where everyone on the call must agree.
  • It stops when asked. “Don’t call again” is honored, every time.

What makes an AI call clearly unlawful or risky

  • Pretending to be a person, or a specific person. Depending on the facts, impersonating a business or government agency breaks the FTC’s impersonation rule (16 CFR 461), and passing an AI off as a human can be a deceptive practice under state law.
  • Calling mobile phones without consent. Each unconsented artificial-voice call to a mobile number can carry statutory damages of $500, or up to $1,500 if willful (47 U.S.C. 227(b)(3)).
  • Faking caller ID to mislead or defraud (47 U.S.C. 227(e)).
  • Campaigns: sales calls, surveys, political calls or debt collection, which each have their own rules.

Who counts as the caller?

When software places a call for a person, it isn’t always clear who “makes” the call under the TCPA. The FCC looks at who decided whether, when and whom to call, and how involved the platform was in the call (FCC 15-72). That’s why careful services keep a record that a person asked for each call, and name both the person and the service in the opener.

What’s still open

  • AI-specific FCC rules. In August 2024 the FCC proposed rules defining “AI-generated calls” and requiring disclosure on them (FCC 24-84). We found no final rule as of Oct 8, 2026.
  • Training on recorded calls. Recording-consent laws were written before AI training. Whether the other person’s “yes” to recording also covers training is a question courts are still working through, which is why Tuvoz asks about both in the same sentence.
  • State laws. More states may pass AI-disclosure laws. Check what applies where you call.

How Tuvoz handles this

Tuvoz builds these rules into the product, and your agent can’t switch them off:

  • The opener names the AI, you and Tuvoz, and the purpose: “Hi! I’m Jane Doe’s AI assistant, calling through Tuvoz about a dinner reservation.”
  • It asks before recording: “Just so you know, Tuvoz records this call, and it may be used to train AI. Other AI companies may get a transcript with names and numbers removed. Is that okay with you?” A no ends the call, and Tuvoz keeps nothing the person said. Our AI voice provider processes the person’s words live to run the call and may keep them for a limited time for abuse monitoring, as our Privacy Policy explains.
  • It never claims to be human or you. A monitor ends the call if the voice agent ever says otherwise.
  • It gives a callback number that isn’t the number calling you: when the voice agent ends the call, including after a no or a request to stop, and in any voicemail. Calling it plays a short message, and pressing 9 stops Tuvoz calls to your number.
  • It checks the number first. Listed business landlines, your own phone, and personal numbers only when the owner agreed and you added them. It can’t call unlisted home landlines yet. No toll-free numbers, emergency lines, crisis lines or premium numbers.
  • It follows state limits. No calls to numbers in Indiana, Louisiana, Minnesota, Montana or Nebraska, the same five states Google’s business-calling feature leaves out. North Dakota: listed businesses only. California: no unlisted home lines. Texas: only once Tuvoz holds the state permit.
  • It calls businesses and other people 9 am to 8 pm in the called number’s time zone. Calls to your own phone aren’t limited to those hours.
  • It keeps a record that you asked: your words, which agent sent the request, and whether you named the business yourself. If your agent picked the number, or the run is unattended, you approve the call in the dashboard first.
  • It honors opt-outs. “Don’t call again” or pressing 9 stops Tuvoz from calling that number again, for every Tuvoz user.
  • It refuses campaigns: telemarketing, political calls, fundraising, debt collection, surveys and lead generation.

None of this makes any particular call legal; it keeps the calls Tuvoz places inside the rules above. You’re still responsible for what you ask your agent to do.

Read next: AI call disclosure rules by state, or how we use call data.

Sources

  1. FCC Declaratory Ruling FCC 24-17: AI-generated voices are 'artificial' under the TCPA (Feb 8, 2024) (accessed Oct 8, 2026)
  2. 47 U.S. Code § 227 (Telephone Consumer Protection Act) (accessed Oct 8, 2026)
  3. 47 CFR § 64.1200 (Delivery restrictions) (accessed Oct 8, 2026)
  4. FCC 15-72: 2015 TCPA Omnibus Declaratory Ruling and Order (accessed Oct 8, 2026)
  5. FCC 24-84: Notice of Proposed Rulemaking on AI-generated calls and texts (Aug 2024) (accessed Oct 8, 2026)
  6. 16 CFR Part 461: Rule on Impersonation of Government and Businesses (accessed Oct 8, 2026)
  7. California Penal Code § 632 (accessed Oct 8, 2026)
  8. California Public Utilities Code § 2874 (as amended by AB 2905, 2024) (accessed Oct 8, 2026)
  9. Minnesota Statutes § 325E.27 (accessed Oct 8, 2026)
  10. Maine Revised Statutes Title 10 § 1500-DD (accessed Oct 8, 2026)
  11. Utah Code § 13-77-103 (Required disclosures) (accessed Oct 8, 2026)
  12. Google Business Profile Help: About automated calls and texts from Google to your business (accessed Oct 8, 2026)