Consent for AI calls

AI calling consent, asked on WhatsApp and kept on file

Two things are checked before your AI agent rings anyone: has this person said YES to your WhatsApp calling consent question, and are they on your do-not-call list? The question goes out under your business’s name, every answer is kept on file, and your other dialers can ask the same list before they ring.

  • One do-not-call list
  • No calls outside 08:00 to 21:00, their time
  • Every reply on file for 5 years at least
How it runs

In order, from first message to first call

Nothing is rung until the first two steps are done. The fourth can happen at any point, and it always wins.

  1. Your business asks

    One WhatsApp message goes out in your business’s name: may we call you about this, and may the call use an AI voice? The person replies YES, NO or STOP.

  2. The answer is filed

    Their exact words are kept with the version of the question they saw, on a record that can only ever be added to.

  3. A call at a sensible hour

    A YES puts the person in the queue for an AI call campaign, and the AI agent rings when it’s within calling hours where they live.

  4. A stop overrides the lot

    A STOP on WhatsApp, or “please don’t ring me again” said on a call, adds the number to the do-not-call list, and any call already queued for it is cancelled.

The question

Your WhatsApp calling consent message, as it’s sent

The message carries your business name and the reason for the calls, and says plainly that an AI voice may be used. It links your privacy page and asks for YES, NO or STOP. Every edit makes a new version, and the old ones are kept.

Say a kitchen showroom in Leicester wants to arrange design visits. It asks the people who requested a brochure on WhatsApp in the last 30 days whether its AI assistant may ring them about it.

  • Who you can ask: people who wrote in recently, a past WhatsApp campaign, pasted lines, or a CSV file of up to 2 MB
  • Skipped without you doing anything: do-not-call numbers, people who’ve already said yes and anyone asked lately
  • Only sent during the person’s own daytime, and never more than twice in 30 days
Replies

YES, NO, STOP and START, one by one

The time limits shown are the defaults. You can change them in the consent settings.

They replyAI callsWhat happens next
YESAllowedGood for 12 months by default. If they reply NO later, it’s withdrawn.
NONot allowedThey aren’t asked again for 180 days by default.
STOPStopped at onceCampaign messages and AI calls stop, anything already queued is cancelled, and from then on the number is on your do-not-call list. Lifting it takes a START from the person themselves.
STARTNot until a new YESMessages are allowed again, but nobody rings until they say YES to a fresh question. A number your team blocked stays blocked.
Calling rules

The rules every call has to pass

Each rule is checked when a person is queued and again just before the call. If anything changes in between, they’re taken back off the queue.

  • A YES to the exact question, given within the last 12 months
  • Between 08:00 and 20:00 local time by default, and never outside 08:00 to 21:00
  • At most 3 tries at a number in any 24 hours, counting every campaign
  • Sundays and any dates you’ve blocked are left alone
  • A number whose time zone isn’t known is never rung

One more condition lives in the phone settings: the person’s country has to be one of your allowed countries for calls, or the call can’t go out.

Evidence

Every answer, sealed in order

Every question, answer and withdrawal goes onto a chain of records that can only be added to. Each record is sealed together with the one before it.

  • Verify goes back over every record, first to last, and says so if one has been edited or deleted
  • The activity log records changes to the settings and the lists, and who made them
  • Open anyone’s timeline to see the question they saw, their exact words and when they replied
  • Kept for 5 years at the very least, and ready to export to CSV whenever someone asks you for proof
Do not call

Three ways a number gets blocked

Once a number’s on this list, the AI agent won’t ring it, and WhatsApp campaigns and the inbox leave it out.

A reply that says STOP

A reply of STOP adds the number at once. It’s understood in many forms, including French, Spanish and Arabic.

Heard during an AI call

“Please don’t ring me again” works the same way, once the do-not-call tool is ticked on the agent.

Entered by hand, with a reason

Your team can add a number and note why. If it’s taken off later, the entry’s history stays on file.

One exception to know about: the WhatsApp tool an agent can use mid-call doesn’t consult this list. Don’t give that tool to the agents that make your AI calls.

Other dialers

Your other dialer can ask the list too

Your other dialer or CRM can put one question to the list before it rings: may I ring this number now? The reply is a plain yes or no with the reason, plus the hours that apply when it’s a yes.

  • Each system gets its own private key, and it only works from the IP addresses you entered for it
  • Make the check at the moment of dialling, not the night before: each yes uses up one of the number’s 3 attempts in 24 hours
  • If someone tells that system they don’t want calls, it can send the number back to go on your do-not-call list
Before you start

What has to be in place first

The app’s overview page keeps this list for you, with a Set up link beside anything that isn’t done yet.

  • Level 2 business verification, done on the main account
  • WhatsApp connected to the account, with the App Secret pasted in
  • The basics filled in: who’s calling, the reason for the calls and where your privacy page lives
  • Meta’s approval for the consent template
  • An agent greeting that admits it’s an AI assistant, names your business and offers people the chance to ask for no more calls
  • The do-not-call tool switched on for that agent, at its Connections step
FAQ

Asking first, in more detail

Not covered here? Send us your question

What’s the point of the WhatsApp question?

Rules on automated and AI-voice calls differ from country to country, and a good many expect the person’s agreement before you ring. Asking on WhatsApp gets that agreement in writing, in their own words, filed next to the wording they were answering.

Can you show me the wording people get?

In plain words, it says who’s asking and why they’d like to ring, warns that the calls may come from an AI voice assistant and makes clear that agreeing doesn’t commit anyone to buy. Then it invites a YES or a NO, says STOP works too, and points to your privacy page. You can use your own approved template instead, as long as it names the business, mentions the AI voice and asks for YES, NO and STOP.

Does agreement run out?

Yes, after 12 months, unless you’ve set something shorter. It can end sooner, too: a NO or a STOP afterwards cancels it on the spot.

Can a person opt out while the agent is talking to them?

Yes, as long as that agent has been given its do-not-call tool. When someone says they don’t want another call, the number is blocked before they hang up, and any calls already lined up for it are cancelled.

We dial from another system too. Can it check the list?

Yes. Create a key for it, and before each call it asks whether that number may be rung now. It gets back a yes or a no with the reason. If someone tells your other system they don’t want calls, it can pass that back and the number goes on your do-not-call list.

Does this replace screening against national registers?

No. It covers your consent records and your in-house do-not-call list, nothing beyond them. Screening numbers against national or state registers is still up to you.

Ask first. It’s the polite way to ring.

Open a free account, link WhatsApp, and tick off everything on the overview page before the agent rings anyone.