How to protect yourself from AI scams: 5 scams, 5 habits

The best protection from AI scams is a habit, not a sharp eye: whenever a call, video, email or message asks for money, codes or secrecy, stop and confirm it through a channel you choose, such as a number you already have. AI can copy a voice, a face and a writing style, but it cannot answer the phone of the person it is pretending to be.

Below are five AI scams the FBI and FTC warn about, one habit for each, and where to report.

How common are AI scams?

Common enough that the FBI now tracks them. Its 2025 Internet Crime Report counted 22,364 complaints that referenced AI, with more than $893 million in reported losses. Investment scams with an AI link were the biggest share, at over $632 million.

The FBI also notes that overall investment scam losses passed $8 billion, which it reads as a sign that "many victims do not realize the extent AI may be involved in scams." So the real number of AI-assisted scams is likely higher than the AI-tagged complaints show.

None of this means AI has invented new crimes. The FBI's December 2024 alert on generative AI describes old schemes, such as fake emergencies, fake investments and fake romances, made faster and more believable. The old defenses still work.

The five scams and the one habit for each

1. Voice clone calls: agree on a code word

A scammer takes a short clip of someone's voice from a video they posted, runs it through a voice-cloning tool, and calls a relative claiming an accident, an arrest or a kidnapping. The FTC described this in its alert on AI family emergency schemes: "All he needs is a short audio clip of your family member's voice."

The FBI counted over $5 million lost in 2025 to these "distress" scams, and warns they are spreading to other family members and close friends.

The habit: set a secret word or phrase with your family that a caller has to know. The FBI recommends exactly this. Then, whatever the caller says, hang up and call the person on the number you already have. The FTC's version is short: "Don't trust the voice. Call the person who supposedly contacted you and verify the story."

For the kidnapping version, the FBI's December 2025 alert on faked proof-of-life photos adds: "Stop and think; do the kidnapper's claims make sense?" and "Always attempt to contact your loved one before considering paying any ransom demand."

2. Deepfake video calls: confirm on a channel you already use

Video used to be proof. The FBI now warns that criminals generate video "for real time video chats with alleged company executives, law enforcement, or other authority figures," and use video to "prove" an online contact is a real person.

The FBI lists visual hints: distorted hands or feet, unrealistic teeth or eyes, irregular faces, odd glasses or jewelry, wrong shadows, watermarks, lag, voice that does not match, and unnatural movement. In job interviews it has seen lip movement that does not line up with the audio, and coughs or sneezes that do not match what is on screen. Treat these as hints, not tests. The FBI itself says AI content "has advanced to the point that it is often difficult to identify."

The habit: if a video call ends in a request for money, passwords or confidential files, confirm it outside that call before you act. Message the person on the chat you already use with them, or call the number in your company directory. A short "can you confirm on Slack?" costs nothing if the call was real.

3. AI phishing emails and texts: never act from the message

AI removed the spelling mistakes. CISA says that "in the era of artificial intelligence (AI) some emails will now have perfect grammar and spelling," on its page about how to recognize and report phishing. The messages can also mention your boss, your bank or a recent order.

The FBI has also tracked campaigns that pair texts with AI-generated voice messages from fake officials, then push the target to move to an encrypted chat app "almost immediately," where they ask for authentication codes, passport copies or wire transfers, according to its December 2025 alert.

The habit: do not sign in, pay or share a code from a link or number inside the message. Type the company's address yourself, or contact the person through a channel you already had. And never read out a two-factor code to anyone. Our guide on how to spot a phishing email shows how to check the real sender and run a quick header check.

Two sign-in upgrades make phishing far less useful to a scammer: passkeys, which cannot be typed into a fake site, and a password manager, which only fills in your password on the correct site, so a lookalike page stands out.

4. Fake AI investment pitches: no one guarantees returns

This is where the money is. The FBI says investment scammers use AI to run thousands of conversations that each feel personal, and that "investment clubs employ AI-generated videos and voices of celebrities, CEOs, or trusted figures to create fraudulent, high-stakes opportunities."

The FTC's page on investment scams lists the signs: promises of "guaranteed profits or returns," claims of little risk and effort, contacts on social media who say they got rich with crypto, celebrity endorsements, and high-pressure sales that discourage research.

The habit: treat "guaranteed" as the end of the conversation. Before you send anything, search the company name with words like "scam" or "complaint," and never move money because a video or a stranger told you to hurry. The FBI's rule fits here too: do not send money, gift cards or cryptocurrency to people you have only met online or over the phone.

5. AI romance scams: never pay someone you have not met

AI gives romance scammers fake profile photos, fluent messages in any language and even short videos to "prove" they are real, according to the FBI. The FBI counted over $19 million lost in 2025 to romance and confidence scams with a likely AI link.

The FTC's romance scam guide describes the pattern: they cannot meet in person, they eventually need money for a crisis or an investment, and they want it by wire, gift card, payment app or crypto.

The habit: "Never send money or gifts to a sweetheart you haven't met in person," as the FTC puts it. If something feels off, talk to a friend or family member, and run a reverse image search on their profile photos.

Make yourself a harder target

A few settings reduce what scammers can work with:

• Share less voice and video in public. The FBI suggests limiting online content of your image or voice, making social accounts private and limiting followers to people you know.
• Turn on two-factor authentication and keep it on. The FBI warns that scammers try to talk people into reading out codes. Our guide on how safe two-factor authentication is compares the methods.
• Verify new numbers. If someone you know contacts you from a new number or app, the FBI says to check it through a previously confirmed platform first. This matters on chat apps too; see how safe WhatsApp is for what end-to-end encryption does and does not protect against.

Where to report an AI scam

Report even if you did not lose money. It helps the agencies spot campaigns.

• Any scam, US: ReportFraud.ftc.gov.
• Online fraud and money lost: the FBI's ic3.gov. The FBI asks for names, numbers and email addresses used, payment dates, amounts, account numbers and crypto addresses, and how the contact started.
• Phishing emails and texts: forward emails to reportphishing@apwg.org and texts to 7726 (SPAM), per the FTC.
• Investment fraud: the SEC at sec.gov/tcr, or the CFTC at cftc.gov/complaint for commodities, as listed by the FTC.
• Identity theft: IdentityTheft.gov gives a recovery plan.
• Money already sent: call your bank or payment service right away and ask whether the payment can be stopped or reversed. Tell the platform where you met the scammer too.

If a scammer got into your email, see what to do if your email is hacked.

Frequently asked questions

How can I tell if a voice on the phone is AI?

Often you cannot. The FBI says cloned voices and real ones "can sound nearly identical." Listen for odd tone or word choice, but rely on the code word and a call back to a number you already have.

What is the most common AI scam?

By money lost, AI-linked investment scams, with over $632 million in reported losses to the FBI in 2025. By number of complaints that mention AI, investment scams also lead the FBI's 2025 list, ahead of extortion and personal data breaches.

What should I do if I already paid?

Contact your bank or payment provider immediately, then report at ReportFraud.ftc.gov and ic3.gov with every detail you have. Be wary of anyone who later offers to recover your money: the FBI has warned about fake law firms that target crypto scam victims this way.

Keep scammers away from your main inbox

Many AI scams start with an email address taken from a data breach or a public profile. With Secria Mail you can give each site its own alias, so if one leaks or starts receiving targeted phishing, you can see exactly where it came from and delete that alias. Tracking pixels are removed automatically and senders never see your IP address. Sign-up is a name, the address you want and a password, with no phone number. Create a free account.