AI calling laws in 2026: consent, disclosure, and what local businesses need to know

For a while, AI voice agents lived in a regulatory grey zone. That's over. In 2026, the rules that have always governed phone calls — and a new wave of AI-specific disclosure laws — apply to AI voices too. The good news: for the most common local-business use case, the rules are manageable. The key is knowing which rules apply to what you're actually doing.

The big one: the FCC says AI voices count as "artificial"

In February 2024, the FCC issued a declaratory ruling confirming that an AI-generated voice is an "artificial or prerecorded voice" under the Telephone Consumer Protection Act (TCPA). In plain terms: a call that uses an AI voice is treated like a robocall.

Under the TCPA, calls using an artificial or prerecorded voice generally require the called party's prior express consent — and for marketing calls, prior express written consent — unless an exemption applies. So the headline is simple: if you're using an AI voice to call people, consent is the foundation.

Inbound vs outbound — the distinction that matters most

This is the single most important thing to understand, and it's where a lot of fear is misplaced.

• Inbound: the customer calls you. When someone dials your business and your AI agent answers, you are not robocalling them — they initiated the contact. This is the core Televox use case, and it carries far less TCPA consent risk than outbound dialing.
• Outbound: you call the customer. Appointment reminders, follow-ups, lead callbacks, campaigns — this is where TCPA consent rules bite hardest, because now your business is the one placing the call with an artificial voice.

Consent can be revoked — and you must honor it fast

A rule that took effect on April 11, 2025 tightened the opt-out side. Consumers can revoke consent through any reasonable means — and you can't force them through a specific script to do it. The FCC spelled out that words like "stop," "quit," "end," "revoke," "opt out," "cancel," or "unsubscribe" clearly count.

Two practical consequences:

• You have to actually process opt-outs — within 10 business days, and across the channel. A revocation on a call can apply to texts, and vice versa.
• Your AI agent needs to recognize an opt-out when it hears one mid-call and act on it, not just sail past it.

The new frontier: AI-disclosure laws

Beyond consent, a second wave of rules is about honesty — telling people they're talking to AI.

• At the federal level, the FCC has proposed (not yet finalized) rules that would require an explicit disclosure that AI is being used, both when getting consent and at the start of an AI call.
• Utah already requires businesses to disclose they're using AI when a consumer asks — and to proactively disclose in higher-risk, regulated interactions (think health, finance, and legal advice). The state has fined for violations.
• California's "bot" law makes it unlawful to use a bot to mislead someone about its artificial identity in order to incentivize a sale. The throughline across states is the same: don't let people believe a machine is a human.

The trend line is unmistakable. Disclosure is moving from "nice to have" to "expected," and the safest posture is to get ahead of it.

A practical checklist for a local business

You don't need a compliance department to get the basics right:

1. Have your agent identify itself. Let callers know they're speaking with an AI assistant for your business. It builds trust and gets ahead of disclosure laws.
2. For outbound, get and keep consent. Don't place AI voice calls to people who haven't agreed to be contacted. Keep records of how and when you got consent.
3. Honor opt-outs immediately and everywhere. "Stop" means stop — on calls and texts, within the required window.
4. Mind the map. State rules differ, and what matters can be where your callers are, not just where you are. If you operate across state lines, get advice.
5. Don't impersonate a human. If a caller would be surprised to learn they were talking to a machine, you haven't disclosed enough.

How we approach this at Televox (honestly)

We build the controls that make the responsible path the easy one: your agent can clearly identify itself as an AI assistant, it can recognize and act on opt-out language, and on the outbound side you control who gets contacted and when. What we won't do is claim a certification we don't hold. We give you the controls; compliance for your specific calling program is ultimately yours to own, ideally with a lawyer who knows your industry.

We write about this the same way we write about everything else: plainly, without overclaiming, and with the parts that are still your responsibility clearly marked.

Frequently asked questions