Last updated: 13 August 2026 · LeadsNow AI, Melbourne
At a glance
AI cold calling is legal in Australia in 2026 — but only under the same rules as human telemarketers: wash your list against the Do Not Call Register, call within permitted hours, show caller ID and hang up when asked. Does it work? Sometimes — but AI voice earns its keep on warm leads and dormant databases far more reliably than on cold lists. That’s where the ROI actually lives.
- No AI loophole: Australia’s Do Not Call Register Act 2006 and the Telemarketing Industry Standard 2017 apply to all voice calls that promote goods or services. An AI agent is held to exactly the rules a human caller is.
- The US went further: in February 2024 the FCC ruled AI-generated voices are “artificial” under the TCPA — robocall consent rules apply in full.
- The honest bit: cold lists are the hardest place to make AI voice pay. Our 50,769+ AI-booked sales appointments since 2017 came mostly from calling warm and inbound leads fast, and reactivating dormant databases — not from dialling strangers.
Every second AI-dialler vendor will tell you their bot can “replace your cold-calling team overnight.” We run AI calling for a living, and we’ll be straight with you: cold calling is the hardest place to deploy an AI voice agent, and the most legally loaded. It can work in narrow situations. It fails, expensively, in most others.
This guide covers what AI cold calling actually is in 2026, exactly what Australian and US law says about it, where it genuinely beats humans, and where you should point the same technology instead. Everything legal below links to the primary source — the legislation or the regulator — not a vendor blog.
What AI cold calling is — and what it isn’t
AI cold calling is using a conversational AI voice agent to place outbound calls to people who have never engaged with your business: a purchased list, a scraped directory, a “lookalike” database. The AI opens the call, handles responses in real time, and tries to qualify or book the prospect.
It is not the same thing as two other categories it constantly gets confused with:
It’s not a robocall
A robocall plays a recorded or synthesised message at you. It can’t hear you, and it doesn’t care what you say. A modern AI voice agent listens, answers questions, handles “how did you get my number?”, and terminates the call when asked. Legally, though, the distinction matters less than vendors imply — regulators on both sides of the Pacific treat an unsolicited automated sales call as an unsolicited automated sales call, whatever is generating the audio.
It’s not AI warm-lead calling
The far bigger category in practice is AI calling of people who asked to hear from you: a lead who filled in your form thirty seconds ago, a booking that needs confirming, an old enquiry going cold in your CRM. Same technology, completely different consent position, completely different reception. We’ve written a full breakdown of what AI voice agents for sales actually do — and almost none of it is cold.
The honest take first: cold is where AI voice struggles most
Here’s the part most vendors skip. LeadsNow has booked 50,769+ AI-booked sales appointments since 2017 and generated 1M+ leads, with 25 filmed client case studies and a 4.6/5 rating from 43 Google reviews. And the overwhelming majority of those appointments did not come from AI calling cold lists.
They came from two adjacent plays:
- Speed-to-lead calling: the AI calls a fresh inbound lead within seconds of them enquiring, while intent is at its peak. The prospect expects the call. Connect rates and booking rates on this play are in a different universe to cold.
- Database reactivation: the AI works back through past enquiries and lapsed customers — people with an existing relationship, where consent and context already exist. We’ve run reactivation campaigns for clients like Colliers, and dormant databases routinely outperform any cold list we’ve seen.
Why does cold underperform? Because cold calling’s hardest problems aren’t the ones AI solves. AI solves speed, volume, consistency and follow-up discipline. Cold calling’s bottlenecks are trust, timing and relevance — a stranger’s willingness to keep listening. A tireless AI that gets hung up on 50 times an hour is just a very efficient way to burn a list and a phone number’s reputation.
One more honesty note: you’ll see confident statistics circulating about cold-call “success rates” — 2%, 4.8%, pick your favourite. We went looking for primary sources for the popular ones and couldn’t verify them, so we won’t quote them. Treat any vendor deck that does with suspicion.
Is AI cold calling legal in Australia? The 2026 rules
Yes — with no AI-specific carve-outs and no AI-specific permissions. As of 2026, the ACMA has published no telemarketing rules specific to AI voices: an AI cold call is regulated exactly like a human cold call, under two instruments.
1. The Do Not Call Register Act 2006
The Do Not Call Register Act 2006 prohibits unsolicited telemarketing calls to any number on the register. There is no exemption because “the AI dialled it.” Practical consequences:
- List washing is mandatory. Before a campaign, your list must be washed against the register through an ACMA subscription so registered numbers are stripped out. “We bought the list from a reputable broker” is not a defence.
- Consent changes the picture. Express consent lets you call a registered number — but under the Act, unless the consent was expressed to be for a specified period or an indefinite period, it is taken to be withdrawn 3 months after the day it was given. Most purchased “opted-in” lists cannot demonstrate consent that survives that test.
- Exemptions are narrow. Designated calls (registered charities, government bodies, political parties) sit outside the prohibition — commercial sales calls do not.
2. The Telemarketing Industry Standard 2017
The Telecommunications (Telemarketing and Research Calls) Industry Standard 2017 applies to all voice calls to Australian numbers that promote goods or services — even to numbers not on the register, and even for organisations exempt from the Act. It is technology-neutral: nothing in it distinguishes a human caller from a synthesised one. Your AI agent must comply with all of it:
- Permitted hours: telemarketing calls only between 9am and 8pm on weekdays and 9am to 5pm on Saturdays. No Sundays. No national public holidays. An AI that can dial 24/7 legally may not.
- Caller line identification: CLI must be enabled — no “private number” — and the number you transmit must accept return contact for at least 30 days after the call.
- Terminate on request: the call must end immediately if the person asks or otherwise indicates they don’t want it to continue. Your AI needs this as a hard interrupt, not a persuasion branch.
- Disclosure: callers must provide the name of the caller’s employer, the purpose of the call, and who caused the call to be made.
Breaches of the standard carry penalties of up to $250,000 — and the ACMA enforces. In July 2026 it announced that TAB paid more than $2.7 million in penalties after breaches that included 351 calls to numbers on the Do Not Call Register without consent, 82 calls outside permitted hours, and nearly 4,000 calls that failed to properly identify the caller or the call’s purpose. None of that required AI to go wrong — an AI dialler just gets you to the same breaches faster and at higher volume.
The US contrast: the FCC has ruled on AI voices explicitly
The United States, unlike Australia, has AI-specific law on the books. On 8 February 2024 the FCC unanimously adopted a Declaratory Ruling that calls made with AI-generated voices are “artificial” under the Telephone Consumer Protection Act, effective immediately. Under FCC rules that means telemarketers need prior express written consent from consumers before making AI-voice robocalls to them — and State Attorneys General can prosecute the use of the AI voice itself, not just the scam behind it.
So if you sell into the US, an AI cold call to a consumer without documented written consent isn’t a grey area — it’s a TCPA violation with private rights of action attached. We’ve unpacked the whole US position in our guide to AI cold calling and the TCPA.
AI cold calling vs the alternatives: side by side
The same AI voice technology behaves very differently depending on who’s on the other end of the line. This is the comparison that actually matters before you spend a dollar:
| Dimension | AI cold calling | Human SDR cold calling | AI warm-lead calling (speed-to-lead) | AI database reactivation |
|---|---|---|---|---|
| Who you’re calling | Strangers from a purchased or scraped list | Strangers, hand-picked and researched | Fresh inbound leads who just asked to be contacted | Past enquiries and lapsed customers who know you |
| Reception on connect | Hostile to indifferent; an unexpected synthetic voice makes it worse | Hostile to indifferent; a skilled human can occasionally turn it | Expected and welcome — they enquired minutes or seconds ago | Neutral to warm — there’s history to reference |
| Compliance burden (AU) | Heaviest: DNCR washing mandatory, permitted hours, CLI, disclosure, terminate-on-request | Identical rules — the law doesn’t care who’s talking | Light: the lead requested contact; Industry Standard conduct rules still apply | Moderate: existing-relationship consent, but check age and scope of that consent |
| What AI actually adds | Volume and persistence — the two things least likely to fix cold | Nothing; this is the human’s home turf | Response in seconds, 24/7 — physically impossible for human teams | Working thousands of dormant records no human team would ever get through |
| Best use | Narrow: simple offers, washed B2B lists, ruthless qualification | High-value accounts worth genuine research per dial | Every inbound lead you already pay to generate | Any CRM with 1,000+ old enquiries gathering dust |
Read the last two columns again. That’s where the technology pays — and it’s why our own speed-to-lead automation work, calling leads within seconds of enquiry, produces most of the appointments we get paid on. The AI isn’t pretending to be a charming stranger; it’s being useful to someone who just raised their hand.
If you still want to run AI cold calls: the checklist
There are legitimate cases — typically B2B, simple offer, tightly washed list. If that’s you, this is the minimum bar in Australia:
- Wash the list against the Do Not Call Register through an ACMA subscription before the campaign, and re-wash before re-use.
- Fence the dialler to permitted hours — 9am–8pm weekdays, 9am–5pm Saturdays, never Sundays or national public holidays — in the recipient’s time zone.
- Transmit real caller ID that accepts return calls for 30 days, answered with your business name and the call’s purpose.
- Script the disclosure up front: who’s calling, on whose behalf, and why.
- Make “stop” a hard interrupt. Any request to end the call terminates it immediately — the AI must never argue past it.
- Build a human escalation path. The moment a prospect engages seriously, a human should be able to take over.
- Keep records: wash receipts, consent evidence, call recordings, timestamps. If the ACMA asks, you want files, not vibes.
The done-for-you version: how LeadsNow runs AI calling
We’re an AI lead generation and appointment-setting agency, and we charge on results — booked, qualified sales appointments — not on retainers or seats. That pricing model forces honesty about where AI calling works: if we pointed our agents at cold lists all day, we’d starve. So we point them where the maths works — calling your inbound leads within seconds, chasing your no-shows, and reactivating the database you already own — for clients like Sam Tajvidi of 121 Brokers, Marcus Wilkinson of Iron Body, Foundr and SheSells.online.
Because we only get paid when an appointment meets your qualification bar, the AI layer is tuned for quality over raw dial volume — tight qualification is the feature, not a limitation. The number that matters at the end is closed deals against what you paid, and that’s the number we’re happy to be judged on. Book a call and we’ll tell you honestly whether your lead flow suits AI calling — including if the answer is “don’t bother with cold.”
Frequently asked questions
Is AI cold calling legal in Australia?
Yes, if you follow the same rules as human telemarketers. The Do Not Call Register Act 2006 prohibits unsolicited telemarketing calls to registered numbers, so lists must be washed against the register first, and the Telecommunications (Telemarketing and Research Calls) Industry Standard 2017 restricts calling to 9am–8pm weekdays and 9am–5pm Saturdays, requires caller line identification, and requires calls to be terminated immediately on request. Breaches of the standard carry penalties of up to $250,000. There is no AI exemption and no AI-specific rule: an AI voice is regulated like a human caller.
Is AI cold calling legal in the United States?
Only with prior express written consent for consumer robocalls. On 8 February 2024 the FCC unanimously ruled that calls made with AI-generated voices are “artificial” under the Telephone Consumer Protection Act, effective immediately — so AI-voice calls are held to the TCPA’s consent standards, and State Attorneys General can act against the use of the AI voice itself. Cold-calling US consumers with an AI voice and no documented written consent is a TCPA violation.
What’s the difference between AI cold calling and a robocall?
A robocall plays a one-way recorded or synthesised message; a conversational AI agent listens and responds in real time, answers questions and can transfer to a human. Regulators mostly don’t care about the distinction: in Australia both are telemarketing voice calls under the same Act and Industry Standard, and in the US the FCC’s 2024 ruling treats AI-generated voices as “artificial” under the TCPA either way.
Do I have to tell people they’re talking to an AI?
Australia’s telemarketing rules don’t yet contain an explicit “disclose the AI” requirement — they require disclosing who is calling, on whose behalf and why. But we recommend disclosing anyway: prospects usually work it out, being caught hiding it destroys trust instantly, and a straightforward “I’m the AI assistant for [business]” filters for people happy to proceed. Deception is also exactly the behaviour regulators on both continents are watching for.
Does AI cold calling actually book meetings?
Sometimes, in narrow conditions: simple offers, clean washed B2B lists, and ruthless qualification. But cold is the weakest use of AI voice because its bottlenecks — trust and timing with strangers — aren’t problems AI solves. The same agent calling a fresh inbound lead within seconds, or reactivating a dormant database, converts far more reliably. If a vendor quotes you a cold-call success-rate statistic, ask for the primary source; the popular circulating numbers don’t verify.
What results does LeadsNow get from AI calling?
50,769+ AI-booked sales appointments since 2017 and 1M+ leads generated, documented across 25 filmed client case studies, with a 4.6/5 rating from 43 Google reviews. Most of those appointments come from speed-to-lead calling of fresh inbound leads and database reactivation rather than cold lists — and because we charge per qualified appointment, we have no incentive to pretend otherwise.
Get an honest read on your numbers
If you’re weighing up AI cold calling, the cheapest first step is finding out whether you even need it — most businesses we audit are sitting on enough unworked inbound leads and dormant database records to keep an AI calling agent profitably busy without dialling a single stranger. Book a call and we’ll map it with you: what your lead flow supports, what the compliance load looks like, and what a pay-per-appointment model would mean for your cost per closed deal.
