AI now makes it cheap and easy to fake a person’s face and voice. Scammers are using these “deepfakes” in calls, Direct Messages, emails and ads to push investment schemes, steal logins, or socially engineer payments.

Why this matters

• Australians reported $2.03 billion in scam losses in 2024.
• Losses were $2.74 billion in 2023.
• On social media specifically, $43.4 million in losses was reported in just Jan–Aug 2024, with thousands of “celebrity-bait” deepfake pages and ads removed under Meta’s FIRE program, including those targeting Australian banks.
• Globally, deepfakes now account for a significant share of biometric fraud, approximately 40% in 2024, highlighting just how convincing synthetic voice and video have become.

Real-world cases

• Celebrity deepfake investment ads. Australia’s consumer watchdog has repeatedly warned about fake news pages and deepfake videos of public figures pushing trading platforms.
• Deepfake video meetings. In a widely reported case, a finance employee in Hong Kong was tricked by a deepfaked CFO and colleagues on a video call into paying approximately US$25 million, illustrating the convincing and coordinated nature of these attacks.
• Bank and exchange impersonation alerts. The AFP and the National Anti-Scam Centre (NASC) have issued warnings about bank impersonation scams and crypto-exchange impostors targeting Australians via text messages, emails, and phone calls.

What deepfakes look and sound like

• Voice cloning: just a few seconds of audio can produce a near-perfect voice. Expect pressure, urgency and requests to move money quickly.
• Video fakes: slick interviews, Zoom calls, or ads where lip-sync is almost perfect, backgrounds look subtly odd, or lighting on a face doesn’t match the room.
• Image fakes: profile pics or proof screenshots with mismatched jewellery, blurred ears/hairlines, or warped text.

Tips

1. Avoid the urgency: Scammers create a sense of panic. Hang up or leave the chat. Call back using a number you recognise (e.g., your bank card number, official website).
2. Verify out of band: If a boss or family member asks for money, call a known number or set a pre-agreed safe word for video calls.
3. Challenge the media: Ask the caller to perform a simple action live, such as turning left or showing today’s date on paper. Watch for unusual lighting, frozen teeth/tongue, out-of-sync blinks, or jerky shadows.
4. Never click payment links in texts: Go directly to your bank app; don’t follow links or numbers supplied in the message.
5. Treat celebrity money ads as scams: ASIC-licensed financial advertisements on major platforms in Australia are moving to stricter verification; if you don’t see clear provenance, assume it’s fake.

Practical prevention

• Use passkeys or app-based 2FA (prefer authenticator apps over SMS where possible).
• Use a password manager and create unique passwords for every account.
• Use PayID namechecking; consider transfer limits and “cooling-off” delays for new payees.
• Keep devices up to date; use built-in password and website warnings.
• Hide your voice and video samples from public profiles where practical; lock down who can direct message or tag you.

Scenarios based on actual scam techniques

Scenario 1: The Urgent Bank Security Call

Characters:

• John, a 52-year-old teacher in Sydney
• Scammer posing as “Mary from his bank” using a cloned voice

What happened

John received a call late at night. The caller, sounding exactly like his bank’s security officer (based on a voice sample lifted from an old radio interview John once did), told him that his account was under cyberattack and he needed to transfer $25,000 into a safe-holding account urgently.

The caller used urgent, fearful language: “We can see criminals draining your account right now.” John, panicked, made the transfer through the link they texted him.

Implications

• John lost $25,000, unrecoverable because he authorised the transaction.
• He spent weeks dealing with ID theft risks after sharing personal details with the caller.
• Emotional stress: loss of sleep, anxiety about financial security.

What could have stopped it

• Stop & breathe: Urgent requests = red flag.
• Verify out-of-band: Call back using the number on the back of the bank card, not the one given in the text.
• Channel check: Banks never ask for transfers via text links or over the phone. 

Scenario 2: The Celebrity Investment Video

Characters:

• Priya, a small business owner in Melbourne
• Scammer running a fake crypto investment ad using a deepfake video of a famous Australian TV presenter

What happened

Priya saw a slick Facebook ad featuring a well-known TV presenter explaining how she “doubled her money” with a new crypto platform. The lip movements and voice were nearly perfect, yet clearly a deepfake.

She clicked through, spoke with support staff, and invested $10,000 via bank transfer, expecting guaranteed returns. The platform vanished after two weeks.

Implications

• Total financial loss.
• Ongoing spam calls targeting Priya for more investments — she was added to a victim list sold on the dark web.
• No legal recourse: the ad originated offshore; complex jurisdictional issues.

What could have stopped it

• Challenge the media: No genuine investment opportunity relies on urgency or secrecy.
• Treat celebrity money ads as scams: ASIC warns that guaranteed returns are a scam.
• Report immediately to Scamwatch and eSafety for ad takedown. 

Scenario 3: The Deepfake “Boss” on Video Call

Characters:

• Li Wei, accounts officer in a Brisbane construction firm
• Scammers impersonating her CEO and two other managers in a deepfaked Zoom call

What happened

Li Wei joined a Zoom call where she saw her CEO and two colleagues asking her to urgently pay $250,000 to a new overseas supplier. The faces blinked, nodded, and spoke naturally — but it was a fully AI-generated video based on real LinkedIn photos and YouTube speeches.

Trusting the “CEO,” she processed the payment. 

Implications

• $250,000 company loss; internal investigation triggered.
• Regulatory reporting obligations under anti-fraud and corporate governance rules.
• Staff morale issues; fear of disciplinary action despite being a victim herself.

What could have stopped it

• Challenge the media: Request a live “safe word” or a unique gesture during video calls.
• Maker-checker control: Payments should require a second verification via a different channel (e.g., phone or SMS to the CEO).
• Incident response drill: Staff need training for deepfake risks in payment authorisation.