Consent & calling rules
On this page
A five-minute read that prevents a genuinely expensive mistake. This is not legal advice — rules differ by country and by industry — but these are the questions you must have answers to before an AI agent calls anybody.
Telecoms and privacy rules vary by jurisdiction and change. If you are calling at volume, in a regulated sector, or across borders, take proper advice for your markets. What follows is what to ask about.
The one principle
Regulators do not treat an AI call as a lesser thing than a human one. Everything that applies to your staff calling customers applies here — usually more strictly, because automated calling can be done at volumes a person cannot.
The practical version: if a member of your team could not lawfully make this call at this hour, neither can an agent.
Consent to call
| Situation | Generally | Watch out for |
|---|---|---|
| Someone just filled in your form asking to be contacted | Strongest position. They gave you the number and asked. | Keep the evidence — which form, when, what it said. |
| An existing customer, about their own booking or account | Usually fine as a service call. | Do not let a service call turn into a sales pitch. That changes its character. |
| A past customer you want to re-engage | Often permitted, with limits on how long and how far. | Time limits, and honouring earlier opt-outs. |
| A purchased or scraped list | The riskiest thing on this page. | You usually cannot demonstrate consent, and “the vendor said it was opt-in” is not a defence. |
Capture consent, do not assume it
If your form has a “you may contact me” checkbox, map it into your Lead Callback trigger and switch on Require consent evidence. Leads with no consent value are then never called — enforcement rather than intention. See Lead Callback.
Calling hours
Permitted hours are set by regulators, and they are measured on the recipient’s local clock, not yours. Calling a Lagos number at 9pm Nairobi time is judged in Lagos.
- Lead Callback enforces this for you. Set the window on the trigger, and it is applied in the lead’s own timezone where you have mapped one. A lead arriving at 3am is held and called when the window opens.
- Campaigns run when you schedule them. There is no automatic window, so choosing lawful hours is your responsibility. Schedule deliberately.
- Weekends and public holidays are restricted in many places. The day-of-week setting on a trigger exists for this.
It is tempting — the lead is hot at midnight and cold by nine. It is also exactly how accounts collect complaints and fines. Let the queue hold them; a 9am call to someone who enquired at midnight still beats every competitor calling on Tuesday.
Opt-outs and do-not-call
Someone asking not to be called again must be honoured immediately, permanently, and across every campaign — not just the one they were on.
If the caller asks not to be contacted again, or says to stop calling: confirm that you will remove them, apologise for the disturbance, and end the call. Do not ask why. Do not offer alternatives. Do not try to keep them on the line.
- Vociply checks numbers against your do-not-call list before dialling a lead callback, and de-duplicates repeat submissions.
- Keeping your own lists clean is still on you — strip opt-outs before uploading a campaign list, every time.
- Cap retries. Three attempts is persistent; six is harassment, and in several jurisdictions it is also a breach.
- Many markets operate a national do-not-call register. Know whether yours does and what it obliges you to do.
Recording notice
Vociply records calls and produces transcripts. Most jurisdictions require notice to the participants; some require consent from all parties.
Where notice is needed, put it in the greeting, where it is heard before anything is said:
"Brightsmile Dental, this is Amara. This call is recorded for quality and training - how can I help?"
This applies to Call IQ as well, where the calls are between two humans — and it covers your own staff, not only your customers. Agree the wording with them before switching a line on.
Disclosing that it is AI
A growing number of jurisdictions require you to say so. Even where it is not required, it is the better choice: people who are told adjust their expectations and stay on the line, while people who work it out mid-call often hang up feeling deceived.
"Hi, you're through to Brightsmile's virtual assistant - how can I help?"
One word does it. Never instruct an agent to deny being an AI if asked directly: in some places that is unlawful, and everywhere it is the kind of thing that ends up screenshotted.
Customer data
- Recordings and transcripts are personal data. They fall under GDPR and equivalent regimes. Know your lawful basis and your retention position.
- Access is account-wide. Anyone on your team can read every transcript. Grant team access accordingly.
- Do not have agents read records back at callers. Using CRM data to greet someone by name is helpful; reciting their history is unsettling and can be a disclosure problem.
- Gate anything financial or medical behind verification. Caller ID is not identity — it is spoofable, and phones get borrowed. The Business Central playbook shows what a properly gated flow looks like.
- Never collect card details on a call. Send a payment link instead.
Vociply’s own certifications, encryption and sub-processor details are on the security page. Data Processing Agreements are available.
Pre-launch checklist
Before an agent calls anyone who is not you or a colleague:
- I can explain, for every number on my list, why I am allowed to call it.
- My calling window is lawful in the recipient’s timezone, not mine.
- The agent has an explicit instruction to accept an opt-out immediately.
- Opt-outs and complainants are stripped from my list.
- Retries are capped at three or fewer.
- My greeting includes recording notice where required.
- My greeting discloses AI where required.
- The agent can transfer to a human, and that number is genuinely answered.
- The agent has an explicit rule never to invent prices, dates or policies.
- Anything financial, medical or account-specific is behind verification.
Ten minutes with a colleague who has not been staring at the prompt catches the “oh, it would definitely say that” problems. Better still, have them listen to three recordings — people hear things in a call that they do not see in a document.
Compliance questions
Is calling a lead seconds after they submit a form legal?
Generally yes, and it is the strongest consent position you can be in — they gave you the number and asked to be contacted. The obligations that remain are calling hours, opt-outs and disclosure.
Do I have to tell callers it is an AI?
In some jurisdictions, yes. Everywhere else it is optional and still advisable. It costs a word in the greeting and reduces hang-ups.
Can I use Vociply for healthcare or financial services?
Yes — both are common. They carry additional obligations around identity, disclosure and data handling, and Enterprise offers HIPAA and SOC 2 compliance packages. Involve your compliance function before launch, not after.
What happens if someone complains about an AI call?
Pull the recording and transcript from Call History — you have a complete record of exactly what was said, which is a considerably better position than a dispute about a human call nobody recorded. Then fix whatever the transcript shows.