Key points
- Criminals use AI to clone voices, fake videos and photos, create fake profiles and write flawless phishing messages.
- The technology changes; the scam pattern doesn’t: urgency + secrecy + an unusual request for money or information.
- Verify through a separate channel you already trust — call back on a known number. Agree a family safe word.
- Report quickly: speed matters for recovering money.
How AI changed scams
Scams used to give themselves away: a strange accent on the phone, a clumsy email full of typos, a profile photo stolen from a stock site. Generative AI removes many of those clues. In a December 2024 public service announcement, the FBI’s Internet Crime Complaint Center (IC3) warned that criminals use generative AI to commit fraud on a larger scale and to make schemes more believable. The announcement described AI-generated text, images, audio and video being used in fraud, including fake social media profiles, cloned voices of relatives and deepfake video calls.
The main AI-powered scams
1. Voice-clone emergency calls
You get a call from a family member’s voice: they’ve been in an accident, arrested or stranded, and need money now — and please don’t tell anyone. The US Federal Trade Commission has warned that scammers can clone a loved one’s voice from a short audio clip, such as one posted online. The voice may be convincing; the request is the giveaway.
2. Deepfake video calls and “CEO fraud”
Employees receive a video or audio call that appears to come from a senior executive asking for an urgent transfer or confidential data. The IC3 announcement notes criminals use AI-generated video to impersonate public figures and authority figures in real-time calls.
3. Fake profiles and romance or investment scams
AI-generated photos and conversational text create believable personas on dating apps and social media, which build trust over weeks before introducing an “investment opportunity” or a request for help.
4. Polished phishing
Emails and texts written with AI look professional, personalised and error-free. They may reference real details about you gathered from social media.
5. Fake documents and images
The IC3 warns of AI-generated identification documents and images used to support fraud — including fake proof of identity, and fabricated photos used to make a crisis or a product seem real.
6. Fake AI apps and tools
Imitation “AI chatbot” apps, browser extensions and websites can steal logins or payment details. Download apps only from official stores and the provider’s own website.
Warning signs (that AI can’t hide)
- Urgency: you must act now, before checking with anyone.
- Secrecy: “don’t tell Mum”, “keep this confidential”.
- Unusual payment: gift cards, cryptocurrency, wire transfers, payment apps, cash couriers.
- Requests for codes or passwords, including one-time verification codes.
- A change of channel or details: “use this new account number”, “my phone broke, use this number”.
- Emotional pressure: fear, panic, romance, excitement about returns.
The UK National Cyber Security Centre’s phishing guidance stresses the same pattern: scammers try to create a sense of urgency, authority or emotion so you act before you think.
Visual and audio clues (useful but unreliable)
Older or cheaper deepfakes may show odd blinking, mismatched lip movement, blurred edges around the face, inconsistent lighting, strange hands, or a flat, slightly robotic voice. But quality improves constantly, so don’t treat the absence of glitches as proof something is real. Verification beats inspection.
How to protect yourself and your family
- Agree a family safe word. The IC3 recommends creating a secret word or phrase with family members to verify identity. Ask for it in any unexpected emergency call.
- Hang up and call back on a number you already have — not one given in the call or message. The FTC gives the same advice for suspected voice-clone calls.
- Verify payment changes through a known contact, especially at work. Require two people to approve unusual transfers.
- Limit public voice and video of yourself and children where practical, and set social media profiles to private.
- Use multi-factor authentication on email, banking and social accounts, and never share one-time codes.
- Talk about it. Older relatives are frequent targets of “grandchild in trouble” calls. A five-minute conversation about safe words and callbacks is powerful protection.
For small businesses
- Write down a payment-verification rule: any new or changed bank details are confirmed by phone using a number on file.
- No urgent transfer requested by video, voice or email is executed without a second check.
- Train staff with real examples, and make it safe to question a “boss”.
- Read our guide to AI for small businesses for safe adoption tips.
If you think you’ve been targeted
- Stop contact and don’t send anything more.
- Call your bank or payment provider immediately if money or card details were involved.
- Change passwords and secure accounts if you shared login details or codes.
- Report it. In the US, report to the FBI’s IC3 and the FTC; in the UK, follow the NCSC’s guidance on reporting; elsewhere, contact local police and your national fraud agency.
- Keep evidence: phone numbers, messages, recordings, transaction details.
- Warn others who may be targeted next, such as family members or colleagues.
Quick checklist to print or share
- Unexpected urgent request for money or codes? Pause.
- Voice or video of someone you know? Ask for the safe word or call them back on a number you already have.
- Bank details changed? Verify by phone using a number on file.
- Pressure to keep it secret? Tell someone you trust.
- Link or attachment you didn’t expect? Don’t click — go to the official website directly.
Share this checklist with older relatives, teenagers and colleagues who handle payments. Scammers rely on people not having heard of these tricks; a short conversation removes that advantage.
Remove this box by adding your content here, or hide all notes with
"showEditorNotes": false in site.config.json.Using AI to check suspicious messages
You can paste a suspicious email (with your personal details removed) into a chatbot and ask: “What are the warning signs that this might be a scam?” It’s a useful second opinion — but never let a chatbot be the final judge. Verify through official channels. Our guide on fact-checking AI answers explains why.
Frequently asked questions
Can scammers really clone a voice from a short clip?
Yes. The US Federal Trade Commission has warned that scammers can use a short audio clip of someone’s voice — for example from a video posted online — to create a convincing clone for fake emergency calls.
What is a family safe word?
A secret word or phrase only your family knows, used to confirm identity in an unexpected urgent call. The FBI’s Internet Crime Complaint Center recommends creating one.
Can I trust deepfake detection tools?
Don’t rely on them alone. Detection is an arms race and tools miss fakes and flag real content. Independent verification — calling back on a known number — is more reliable.
I sent money to a scammer. What now?
Contact your bank or payment provider immediately, report to the police and your national fraud agency (in the US, the FBI’s IC3 and the FTC; in the UK, Action Fraud or your national equivalent), and keep all evidence.
Are AI-written phishing emails easier to miss?
Often, because spelling and grammar mistakes — a classic warning sign — are now easy to avoid. Focus instead on the request: urgency, secrecy, payments, passwords or links.
Can AI clone a voice from a voicemail greeting?
Short clips can be enough, according to consumer-protection warnings. Keeping public voice and video limited reduces the material available, but verification habits are the real protection.
Sources
- FBI IC3 — Criminals Use Generative Artificial Intelligence to Facilitate Financial Fraud (PSA, Dec 3, 2024) (opens in a new tab)
- FTC Consumer Advice — Scammers use AI to enhance their family emergency schemes (opens in a new tab)
- UK NCSC — Phishing scams: how to spot and report them (opens in a new tab)
- NIST — AI Risk Management Framework (opens in a new tab)
Facts, prices and features were checked against these sources on November 19, 2026. AI products change quickly; if you spot something out of date, please tell us.