“That sounded exactly like my daughter”: when AI voice-clone scams become fraud, blackmail or carriage-service charges in NSW

A lot of criminal-law topics only really matter once police have already laid charges. This one is different. AI voice-clone scams are now frightening ordinary people before they ever think they need a lawyer. A parent gets a call from what sounds exactly like their son. A business owner gets a voice note that sounds like the boss asking for an urgent transfer. An ex-partner gets a fake voicemail, allegedly from someone they know, demanding money or threatening exposure. For the victim, it feels real. For the person accused, the surprise comes later: what looked like “just a scam call” can quickly become a very serious brief involving fraud, blackmail, using a carriage service to menace, harass or cause offence, and sometimes wider dishonesty allegations.

Scamwatch now warns that scammers can use only a few seconds of a voice recording to create a convincing AI clone, then use that clone to pose as a family member or friend in distress. The ABA and banking sector have also been publicly warning Australians about AI scams involving voice cloning, deepfake videos and generated phishing content. That means this is no longer a niche “future of crime” story. It is a live Australian fraud problem happening through ordinary phones and messaging apps right now.

Why this topic matters now

The reason this issue has become so commercially and legally important is that it combines three things that make criminal cases both persuasive and dangerous: emotional pressure, technological realism, and speed. Scamwatch says voice clones are being used to make emergency appeals feel authentic – the classic “I’ve been in an accident”, “I’ve lost my wallet”, or “I’m in trouble and need money now” scenario. Because the cloned voice feels familiar, victims often act before they verify anything. That creates a perfect bridge between cultural panic and very old criminal law: deception, intimidation, financial transfer, and digital evidence.

It also helps explain why police and prosecutors do not see these cases as “just technology issues”. In legal terms, the core question is still the old one: was money, property or some other advantage obtained by deception, and was the conduct dishonest? s 192E of the Crimes Act 1900 (NSW) makes it an offence where a person, by any deception, dishonestly obtains property, obtains a financial advantage, or causes financial disadvantage. The Commission is explicit that “deception” and “dishonestly” are separate elements the prosecution must prove. That is especially useful in AI-voice cases, because the fake voice is not the offence on its own – it is the mechanism by which deception may be proved.

What is the actual NSW offence if money is obtained?

The starting point in NSW is usually fraud under s 192E. The section is broad enough to cover very modern conduct. The fake voice may be new, but the offence wording is old-fashioned in the best possible sense: if by deception and dishonesty a person obtains a financial advantage or causes another person financial disadvantage, the charge may be made out. The maximum penalty is 10 years’ imprisonment. The Judicial Commission’s sentencing materials also emphasise that courts look closely at amount involved, planning, duration, sophistication, breach of trust and victim impact when sentencing fraud offences. That makes AI voice-clone cases particularly serious because they are often designed to exploit trust directly – family trust, workplace trust, or intimate trust.

For a reader in trouble, that matters because they may still be thinking in technological rather than legal terms. They may ask, “Can I really be charged just because I used AI?” That is the wrong question. The legal question is whether AI was used to deceive someone into parting with money or conferring a financial advantage, or whether the call or message was used to threaten, pressure or blackmail them. The law does not need a separate “AI voice offence” to take this conduct seriously. Existing fraud and telecommunications offences already do a great deal of work.

And what if no money changed hands?

That is where the topic becomes even more practical.

A lot of these incidents do not end with a completed transfer. Sometimes the call is a failed attempt. Sometimes the fake voice is used to frighten, menace, harass or manipulate rather than to obtain money immediately. Sometimes it is used in a personal context: an ex-partner is sent a fake voice note, a family member is targeted with a threatening cloned call, or someone receives repeated false contact designed to terrify or shame them. In those situations, Commonwealth telecommunications law may become central. The Criminal Code Act 1995 (Cth) includes s 474.17, which makes it an offence to use a carriage service in a way reasonable persons would regard as menacing, harassing or offensive. The offence carries a maximum penalty of 5 years’ imprisonment following recent increases noted in legislative materials. The Judicial Commission’s Local Court Commonwealth Offences chapter specifically identifies s 474.17 as one of the offences commonly seen in the Local Court.

That is important because many readers still think telecommunications offences are limited to old-fashioned abusive texts or prank calls. They are not. The explanatory material to the carriage-service offence says it covers conduct judged objectively, including both the content of the communication and the method of use – which is highly relevant to AI voice cloning. A message need not be physically threatening in the old sense to become legally menacing if it is crafted to create fear, urgency, humiliation or psychological pressure.

How police are likely to look at an AI voice-clone allegation

From a practical defence perspective, these cases are likely to be built in very ordinary ways. Police will usually care less about the glamour of the AI than about the evidence trail it leaves: bank transfers, call metadata, app accounts, voice files, chat logs, cryptocurrency movement, device contents, and any admissions. If the call was part of a broader scam, police may also look at fake identities, phishing pages, mule accounts, or other people involved in the chain. That is one reason these matters can widen quickly from a single call into a broader fraud narrative.

The judicial guidance on fraud is useful here because it reminds us that deception and dishonesty are distinct issues. A defence may turn on whether the accused actually created or sent the voice-clone communication, whether they intended anyone to act on it, whether the recipient’s response was induced by the deception, or whether the prosecution can really prove authorship and intent beyond reasonable doubt. In other words, even in a highly modern case, the battleground is still evidence, state of mind, and the prosecution’s ability to join the dots.

What about family and relationship contexts?

This is where the topic gets even more reader-interesting.

Not every AI voice-clone case looks like a pure “bank transfer” scam. Some arise in breakdowns of trust: a fake voice message from a former partner, a cloned audio clip sent to a family member, a message used to trigger fear or public embarrassment. Once the matter moves into repeated unwanted contact, harassment, blackmail, or a domestic setting, the criminal exposure can broaden significantly. In the right case, an AI-generated message could overlap with stalking or intimidation, AVO issues, and carriage-service offending. That is exactly why the reader should not think only in terms of “scam law”. The first question is often: what offence are police likely to choose?

This is where National Criminal Lawyers® can helpfully link our own pages such as Use Carriage Service to Menace, Harass or Cause Offence, What Happens After Charges Are Laid?, and Police Interviews – What Should You Do? Those are useful internal links because, in practice, a reader facing an AI-voice allegation needs immediate procedural guidance as much as substantive law. But the legal backbone of the article should still come from the statute, the Judicial Commission, Scamwatch and recent official warnings – exactly as here.

What should a worried reader actually do if police get in touch?

This is the practical heart of the article.

If you are accused of involvement in an AI voice-clone scam, the biggest mistake is often trying to “clear it up” informally. A person may think the technology sounds so strange that police will obviously understand there has been a misunderstanding. In reality, these cases are often treated as just another fraud or carriage-service matter. The voice-clone element may make the case more alarming to a victim, but legally it still sits inside familiar criminal categories. That means early admissions, half-explanations, and panic messages to other people in the chain can do enormous damage.

Equally, victims should not assume there is no remedy just because the voice was fake. The criminal law is concerned with the deception, the pressure, the use of a carriage service, and any financial or psychological harm caused. If money has moved, fraud issues arise. If threats or menacing conduct are involved, carriage-service offences may be in play. If there is an ongoing campaign of manipulation or humiliation, related NSW offences may follow. The legal system already has tools for these cases; the real challenge is using them quickly and coherently.

Contact Us

If you are being investigated over an AI voice-clone scam, a fake distress call, or any allegation involving a deepfake audio message, fraud, or use of a carriage service to menace, harass or cause offence, do not assume the novelty of the technology will make the case easier. The underlying criminal law is already there, and police will usually treat the matter as seriously as any other deception or intimidation case.

At National Criminal Lawyers®, we can help you understand what offence police are really looking at, what they still need to prove, how digital evidence may be used, and how to avoid making the situation worse in the earliest stage of an investigation.

Call us at 1800 CRIM LAW for a free consultation. Help us help you understand your situation and get you the best outcome possible.