What are AI scams?

An AI scam is a fraud in which the criminal uses artificial intelligence to make an approach more believable or to run it at a scale that was impossible a few years ago. The crime underneath is familiar: a fake relative asking for money, a fake bank alert, a fake government agent, or a fake investment adviser. What has changed is the disguise.

Where a scam call once sounded like a call centre, it can now sound like your grandson. Where a phishing email used to be full of spelling mistakes, it can now read better than most legitimate mail. Where a fake profile used to be an obvious stock photo, it can now be a moving, speaking person.

The fraud is old. The costume is new.

FBI data for 2024 shows that people aged 60 and over lost more money to fraud than any other age group, and that the average loss among victims who reported was far higher than for younger adults. The reason is not naivety. It is arithmetic. Older adults often own a home, hold retirement savings, and have decades of accumulated credit history, which makes them valuable targets. Some also live alone or manage money without a second pair of eyes.

Meanwhile, a younger adult reporting a $400 marketplace loss and an older adult reporting a $40,000 retirement loss appear in the same statistics, but they are not the same event. Scammers know this and price their effort accordingly.

Close-up of an older adult’s hands holding a smartphone that shows a blurred conversation, in a softly lit living room.
Scam messages are written to look ordinary. The warning signs are usually in the timing and the payment request, not in the grammar.

Why AI makes fraud cheaper

Three things used to limit scammers. They needed to speak the language well, they needed to sound like the person they claimed to be, and they needed enough time to work each victim by hand. AI removes all three limits at once.

  • Writing is free. A chatbot can produce a convincing message in flawless English in seconds, then personalise it with details drawn from social media.
  • Voice is copyable. Commercial voice-cloning tools can build a usable imitation from a short, clear recording, such as a voicemail greeting.
  • Time is automated. One operator can run thousands of text conversations at the same time, and hand the promising ones to a person.

Do not treat personal details as proof

A caller who knows your name, your address, your bank’s name, or a family member’s nickname has probably simply bought that information. Large data breaches have exposed billions of records, and stolen customer lists are traded cheaply. Knowledge of your details proves that someone has data about you, not that they are who they say they are.

AI is also part of the defence

It helps to keep the picture balanced. The same technology that powers these scams also powers the spam filters that keep most of them out of your inbox, the fraud-detection systems that flag unusual card activity, and the voice-authentication tools used by some banks. The risk is real, but you are not undefended.

How AI scams work: five techniques in use today

Almost every AI-assisted fraud you are likely to meet uses one of the five methods below. Recognising the method takes the surprise out of the moment.

1. Voice cloning: the relative in trouble

This is the best-known AI scam. The caller sounds like your child or grandchild, often crying or panicked, and explains that they have been arrested, hurt in a car accident, or are stuck abroad. A second person takes the phone, claims to be a police officer or a lawyer, and insists that money must be sent immediately and that nobody else can be told.

Voice-cloning tools need only a short sample of speech, and social media is full of clear recordings of people talking. The emotional shock does the rest: the part of your mind that would normally ask questions is busy worrying about someone you love.

2. Deepfake video and fake video calls

Fake video is used in two ways. In a video call, a criminal uses software to wear someone else’s face in real time, which makes an investment pitch or a request for a wire transfer feel authentic. In advertising, a synthetic version of a famous doctor, actor, or news anchor appears to endorse a product or an investment platform that does not exist.

Treat any famous face selling an investment, a supplement, or a "claim service" as fake until proven otherwise. Major news organisations and public figures are frequently misused this way, and they almost never endorse financial products in online adverts.

3. AI-written phishing messages

Phishing means a message designed to make you click a link, open an attachment, or reply with information. AI makes these messages fluent, correctly punctuated, and personalised with details scraped from your social media. The classic tells — broken grammar and generic greetings — are becoming less reliable, so pay more attention to the request than to the wording.

4. Chatbot romance and investment scams

Long-running confidence schemes now begin with a chatbot that keeps a conversation going for weeks. A friendly stranger meets you on social media or a dating app, shares photos and stories, and eventually introduces an investment "opportunity" that shows impressive early returns on a website the criminals control. When you try to withdraw, you are told to pay tax or fees first.

5. Automated robocalls and fake support agents

AI voices make it cheap to record thousands of calls that respond to what you say. Common versions claim to be from Medicare, Social Security, your bank, or a well-known software company warning of a virus. The goal is a conversation that ends with remote access to your computer, a payment by gift card, or a transfer to a "safe account".

A useful phrase to keep ready

“I don’t handle this over the phone. I’ll call back on the number I have.” You do not need to explain, argue, or be polite about it. A genuine organisation will accept that sentence without pressure. A scammer will try to talk you out of it, which is the answer you need.

Six red flags that mean stop

You do not have to spot the AI. You only have to notice the pattern. These six signals appear in almost every AI-assisted scam, whatever the story around them.

  1. Urgency plus secrecy

    Act now, don’t tell anyone, stay on the line. Real banks, police departments and government agencies never ask you to keep a payment secret.

  2. An unusual payment method

    Gift cards, cryptocurrency, wire transfers, payment apps, or cash collected by a courier. These are chosen because they are hard to reverse.

  3. A request for remote access

    Microsoft, Apple, your internet provider or your bank will not cold-call you and ask you to install AnyDesk, TeamViewer or similar software.

  4. An offer too good to be true

    Guaranteed returns, a prize you never entered, or a job that pays well for very little work. Nothing legitimate depends on you deciding today.

  5. A new number and a new photo

    A message from an unknown number claiming to be family, or a profile picture that does not match the account you know. Call the saved number instead.

  6. Small errors that add up

    Odd wording, a mismatched email domain, a greeting that does not sound like the person, or a logo that is slightly wrong. Check the domain, not the display name.

If a message creates urgency, demands secrecy, or asks for a payment you cannot reverse, stop. Those three things are the scam, regardless of the story attached to them.

Four scenarios you might recognise

Each of the examples below is based on a pattern that fraud investigators see repeatedly. Reading them in advance is useful: recognition is faster than reasoning when you are startled.

Voice cloning

“Grandma, I’m in trouble — please don’t tell Mom.”

What happens
A call comes from an unknown number. The voice is your grandson’s, and he is crying. He says he hit a parked car and was arrested. A second person, calm and official, explains that bail must be paid in cash or gift cards within the hour.
Why it works
The voice is familiar, the emotion is real-sounding, and the request for secrecy prevents you from checking with his parents.
What to do
Hang up. Call your grandson on the number you have saved, then call his parents. If you cannot reach them, ask the caller for your family code word. Courts and police never accept gift cards.
Bank impersonation

“We’ve blocked a $480 charge — confirm your password.”

What happens
A text or call appears to come from your bank’s number. The agent reads part of your account number, warns of a fraudulent charge, and offers to move your money to a "safe account" while the problem is fixed.
Why it works
Caller ID can be spoofed, and partial account details are easy to obtain. The idea of a safe account sounds responsible and official.
What to do
Never read back a password or one-time code. Hang up and call the number printed on your card. Real banks do not move your money to a different account to keep it safe.
Government impostor

“There is a warrant in your name. Pay now and it goes away.”

What happens
The caller claims to be from Social Security, the IRS or a federal law enforcement agency. Your Social Security number has supposedly been used in a crime, and you must withdraw cash to protect it. You are told to stay on the phone while you go to the bank.
Why it works
Authority, fear, and a sense that time is running out. The instruction to stay on the line prevents anyone at the bank from warning you.
What to do
Hang up. Federal agencies do not accept payment by gift card, wire, or courier, and they do not resolve warrants over the telephone. Report the call at ReportFraud.ftc.gov.
Chatbot confidence scheme

“I’ve been watching your progress. You could retire a year earlier.”

What happens
A friendly stranger met online shares a personal story, then mentions a trading platform a relative designed. Early withdrawals work, so you invest more. When you ask for a large withdrawal, you are told to pay tax first.
Why it works
The relationship feels genuine, the first small success is engineered, and the request to pay fees before withdrawing sounds like normal paperwork.
What to do
Stop sending money. Check any investment seller in your state’s securities regulator database or on Investor.gov before investing, and ask a family member to look at the platform with you.

How to protect yourself: seven practical steps

None of these steps requires new technology or an hour of setup. Done once, they cover a very large share of the scams in circulation.

  1. Step one

    Agree on a callback rule

    Make it a household rule that money requests are confirmed by calling back on a number you already have. Not the number on the message, not the number the caller gives you.

    Add the family members you would call to your phone’s favourites so the correct number is always one tap away.

  2. Step two

    Choose a family code word

    Pick a word that only your family knows and that would never appear on social media, such as the name of a childhood pet nobody has mentioned online. Use it when a caller claims to be a relative in trouble.

  3. Step three

    Turn on two-factor authentication

    Add a second step to your email, bank, and social media logins, ideally with an app or a security key rather than text messages. This is the single most effective protection if a password is stolen.

  4. Step four

    Use a password manager

    Long, unique passwords for every account, stored in one encrypted place, are safer than variations of one word you can remember. Write the master password on paper and keep it somewhere private, not in your phone.

  5. Step five

    Set up bank alerts

    Ask your bank for text or email alerts on any transaction above an amount you choose. An early warning is what turns a fraud into an inconvenience instead of a loss.

  6. Step six

    Review what you share publicly

    Public posts about travel, grandchildren, and health give scammers the details that make an impersonation believable. Set your profiles to friends-only and remove visible phone numbers and birthdays.

  7. Step seven

    Keep devices updated

    Install operating system and browser updates when prompted. Updated software will not stop a persuasive lie, but it does close the technical gaps that scams use to deliver malware.

Freeze your credit before anything happens

A credit freeze is free in every U.S. state and takes about twenty minutes to set up with each of the three major credit bureaus. It prevents a criminal who has your Social Security number from opening new accounts, and you can lift it temporarily whenever you need to apply for credit.

How families can help without taking over

Most older adults who are targeted never mention it. Embarrassment is powerful, and scammers deliberately build shame into their scripts so that victims stay quiet. The most protective thing a family can do is make the conversation normal and non-judgmental long before anything happens.

A man in his seventies and a woman in her forties look at a smartphone together at a dining table while a teenager works on a laptop behind them.
Looking at a suspicious message together takes two minutes and settles most doubts.

Five things that help

  • Agree the rule in advance. Decide together that no payment, transfer, or code will ever be sent without a callback. Frame it as a family policy, not as a restriction on one person.
  • Ask about the attempt, not the outcome. Questions like “Has anyone called you claiming to be me?” are easier to answer than “Did you fall for something?”.
  • Offer to be the second pair of eyes. Make it known that a forwarded screenshot will get a fast, calm reply rather than a lecture.
  • Set up the practical safeguards together. Sit beside your relative while they enable two-factor authentication and bank alerts, so the logins remain theirs alone.
  • Keep money and identity separate from help. If support is needed with bills, use a formal arrangement such as a bank’s trusted-contact feature or a power of attorney drawn up properly, not informal access to accounts.

“I’d rather check a message ten times a year than help sort out one transfer. Can you add me to the list of people you call before you send money?”

A sentence that opens the conversation without judgement

If you think you have been scammed

Act in the first hours. Recovery is never guaranteed, but the speed of your first phone call matters more than anything else you do afterwards.

The first 24 hours

  1. Call your bank or card issuer. Use the number on the back of your card. Explain what happened and ask them to stop or recall the payment, freeze the account, and issue new cards.
  2. Secure your accounts. Change passwords for email, banking, and anything reusing the same password. Revoke remote-access software if you installed it, and turn off any forwarding rules you did not create.
  3. Report it. File at ReportFraud.ftc.gov, and at IdentityTheft.gov if personal information was shared. Internet-enabled crime goes to ic3.gov. If cash was collected in person, contact your local police.
  4. Protect your credit. Ask the three credit bureaus for a free fraud alert or a credit freeze. Check your credit reports for accounts you do not recognise.
  5. Tell someone you trust. Secrecy is what lets a scam continue. Telling one person you trust also protects you from the follow-up "recovery" scams described below.

Where to report and get help

  • Fraud and identity theft

    The FTC’s reporting tools produce a personal recovery plan and feed patterns to investigators. Both services are free.

    ReportFraud.ftc.gov · IdentityTheft.gov

  • Online crime, including voice cloning

    The FBI’s Internet Crime Complaint Center collects reports of internet-enabled fraud, deepfakes and impersonation.

    ic3.gov

  • Unwanted calls and texts

    Add your number to the National Do Not Call Registry, and report unwanted sales calls there. Forward spam texts to 7726.

    DoNotCall.gov

  • Investment offers

    Check whether a seller or platform is registered before you send money, and report suspicious offers to your state securities regulator.

    Investor.gov

  • Support and counselling

    The AARP Fraud Watch Network offers a free helpline staffed by trained volunteers, and the Eldercare Locator can connect you to local services.

    AARP Fraud Watch Network · Eldercare Locator

  • Legal or financial advice

    Your state attorney general’s consumer protection office and a licensed financial adviser can help you decide what to do next. Neither should charge you to report a scam.

    Find your state attorney general

Beware the second scam

Shortly after a loss, people are contacted by "recovery specialists", fake government refund offices, and law firms promising to retrieve the money for an upfront fee. Legitimate agencies do not demand payment in advance to return stolen funds, and no one can guarantee a recovery. Treat any such offer as a new fraud.

Reporting routes and agency names change over time. Always confirm current contact details on the official website before sharing personal information. This page is general education, not legal or financial advice.

AI scam or genuine message?

Genuine organisations do occasionally need to contact you. The difference is rarely in how polished the message looks. It is in what the message asks you to do.

Comparison based on guidance published by the FTC and the FBI’s Internet Crime Complaint Center.
What you notice Typical of a scam Typical of a genuine contact
Timing Pay within minutes, or lose the account, the money, or your freedom. A stated deadline you can verify, usually measured in days, with a reference number.
Payment method Gift cards, crypto, wire transfer, cash courier, or a payment app. Your usual method, through a channel you can confirm independently.
Secrecy Do not tell family, do not hang up, do not speak to a teller. You are encouraged to call back on an official number and involve whoever you wish.
Security codes Read me the six-digit code that just arrived. You are told never to share a code with anyone, including staff.
Contact details New number, mismatched domain, a link with an odd ending. Consistent domain, a public number you can look up, and no pressure to click.

Test yourself: can you spot the scam?

Five short cases, no sign-up required. Nothing you choose is stored or sent anywhere.

Question 1 of 5

Loading question…

Frequently asked questions

What exactly is an AI scam?

An AI scam is an ordinary fraud in which the criminal uses artificial intelligence to look or sound more convincing. The tools include voice cloning, fake video, and chatbots that write fluent messages. The goal is always the same: to get you to send money, share a security code, or hand over remote access to your device.

Can someone really copy my voice with a short recording?

Yes. Several commercial services can build a passable imitation of a voice from a few seconds of clear speech, such as a voicemail greeting or a video on social media. That is why a familiar voice on the phone is no longer proof of identity. Use a callback to a saved number or a family code word instead.

What should I do if I already gave a scammer a gift card number?

Call the card issuer immediately using the number on the back of the card or receipt, report the card as used in a scam, and keep the card and receipts. Then report the incident at ReportFraud.ftc.gov. Speed matters because funds on a redeemed card are rarely recoverable.

Will my bank ever call and ask for a one-time code?

No. Banks and other legitimate organisations do not ask you to read back a one-time passcode, PIN, or password. Those codes are designed to prove that you are the person making the transaction. If someone asks for one, end the call and phone your bank on the number printed on your card.

Where do I report an AI voice-cloning call?

Report internet-enabled crime, including voice cloning, to the FBI Internet Crime Complaint Center at ic3.gov. Report fraud and identity theft to the FTC at ReportFraud.ftc.gov or IdentityTheft.gov. If money was taken, contact your bank first, then your local police department.

Is antivirus software enough to stop AI scams?

No. Most AI-assisted scams rely on conversation rather than malicious software, so antivirus alone cannot detect them. Keep your devices updated, use two-factor authentication on important accounts, and treat any unexpected request for money or codes as unverified, whatever the software says.

Sources and further reading

Statistics on this page come from published government reports. Links open the primary sources so you can check them yourself.

External links are provided for reference. We do not control third-party websites and are not responsible for their content or privacy practices.

Written and reviewed by the Senior Fraud Guide editorial team

We summarise published guidance from the FTC, the FBI and the CFPB in everyday language, and we link to the primary documents so you can check every figure. We do not sell scam-recovery services, and we do not accept payment for coverage. Page last reviewed November 3, 2026.