AI cyber safety

How Australian staff can spot AI scams and deepfakes

Published 8 November 2026 · Sources last checked 5 November 2026

You can't reliably spot an AI scam by how it looks or sounds. Scamwatch says scammers use AI to make scams "more convincing, personalised and difficult to detect". The safest habit is to check the request another way, such as by phoning a number you already trust, before you pay, share or click.

How are scammers using AI against workers?

Scamwatch, run by the ACCC, says scammers use technology and AI to make scams more convincing and harder to spot. It says they can shift tactics or offer fake "evidence" when you raise a concern. It also says AI chatbots can run many conversations at once.

ASD's Australian Cyber Security Centre (ACSC) makes a similar point. It says generative AI can allow malicious actors to easily create convincing scams.

So don't rely on a message looking sloppy to catch it out. A polished email or a smooth phone call proves nothing.

The scale is large. The National Anti-Scam Centre's Targeting Scams report for 2025, released on 30 March 2026, recorded $2.18 billion in combined losses and 481,523 scam reports.

What does a phishing message look like now?

Phishing is still the most reported scam. The ACCC says it was the most reported scam type to Scamwatch in 2025, with 65,361 reports. The National Anti-Scam Centre's report puts phishing losses for 2025 at $97.6 million.

Scamwatch lists these warning signs.

  • The message claims to be from a business you deal with, such as your bank.
  • It pressures you to act straight away.
  • The sender details don't match the real organisation.
  • The web address is slightly different from the usual one.

Don't click the link. Go to the organisation's official website or app yourself. Never hand over passwords, card details or one-time codes to someone who contacted you first.

Can I trust a voice or video call from my boss?

Not on its own. Scamwatch says scammers can make fake audio and video that look and sound real. It says a clone can be built from just a few seconds of someone's voice, then used to call you pretending to be a friend or family member in distress. A familiar voice at work is no more proof than one at home.

eSafety describes a deepfake as a photo, video or sound file of a real person, made with AI to show them doing or saying something they didn't. It says detection tools are lagging behind the technology. It suggests looking for blurring around the mouth, eyes and neck, skin that looks inconsistent, and sound that is badly synced. It also suggests asking whether the content is what you would expect that person to say or do.

The fix is simple. If a call, voice note or video message asks for money, gift cards, passwords or a rush job, hang up. Then call the person back on a number you already have. Scamwatch gives the same advice for family and friends.

How do fake invoices and payment redirection work?

Payment redirection was the second biggest scam type by loss in 2025, at $166.8 million. That was up from $152.6 million in 2024, according to the Targeting Scams report.

The ACCC says fake invoices can copy a business's logo and ABN. It says the email can look like it came from a supplier you already deal with, with only the bank details changed.

ASD's ACSC describes business email compromise as a form of targeted phishing. Criminals pose as business representatives or use an employee's hijacked email account. The ACSC lists warning signs such as emails sent from your account that you didn't write, odd login alerts, and emails that have gone missing.

The ACCC urges you to check payment details directly with the business before paying an emailed invoice. Use a number you found yourself and not one from the email. If you work in accounts, ask your organisation whether a second person should approve any change of bank details.

Is it safe to use AI tools at work?

Mostly, if you treat them like any other online service. ASD's guidance on engaging with AI says to apply the same basic security principles you would use for any online service. It says to look at the tool's reputation and privacy policy, and to check whether the output is accurate before you rely on it.

ASD also suggests asking one question before you paste anything in. Do I need to share this information with the AI system? Check your organisation's policy first. Keep client details, passwords and private business information out of tools your workplace hasn't approved.

What if I think I've been scammed?

Act fast. Scamwatch says to contact your bank or card provider straight away, change your passwords, and report the scam to Scamwatch. IDCARE offers free help on 1800 595 160. The ACCC also says to report compromised systems to cyber.gov.au. Tell your manager or IT team too, so others can be warned.

What can I do this week?

  • Pick a rule for yourself. Any request for money, codes or bank detail changes gets checked by phone on a known number.
  • Save the numbers of your key suppliers and your manager somewhere safe, so you don't rely on numbers in emails.
  • Slow down when a message says "urgent". Urgency is the scammer's friend.
  • Ask your manager what your workplace's AI tool policy says.
  • Share this with one colleague, especially anyone who pays invoices.

Want to build safer AI habits across your team? Our free course, Responsible AI for Australian Workplaces, is at aiworkready.com.au. Organisations can find out more on the For Organisations page.

This is general information, not legal advice. Check your organisation's policy.

Sources

Learn the habit, free

Responsible AI for Australian Workplaces teaches what never goes into an AI tool, how to ask well and how to check the answer. Short online lessons, free for individuals.