Alpha Vault · The Receipts Index

When You Can't Trust the Voice

A calm guide to AI voice clones and deepfakes, and the one habit that beats all of them

Real numbers. No hype. Receipts.

Before we start · Seeing and hearing stopped being proof

For your whole life, a familiar voice on the phone meant it was really them, and a video of a person meant they really said it. That is quietly no longer true. Cheap tools can now copy a voice from a few seconds of audio, and can fake a face on a video call. This is unsettling, and it is also manageable, because although the fakes are new, the scams underneath them are old, and one simple habit defeats every one of them. This short guide gives you that habit, and hands it to you in a form you can teach the people you love.

There is nothing to buy here, and nobody to fear more than is useful. The goal is calm and specific: understand the handful of ways this is used against people, and install the one check that makes you and your family very hard to fool.

One · The new costume on an old machine

The fake is new, the con is not

An AI-cloned voice or a deepfake video is not a new kind of scam. It is a new mask on scams that have run for decades: the loved one in sudden trouble, the boss with an urgent order, the famous name endorsing an investment. What the fake adds is the removal of your natural doubt. When the voice is your daughter's, or the face is your CEO's, the part of you that would normally pause and check goes quiet. That silencing of your doubt is the entire product. Everything else is theater.

So the defense is not to become an expert at spotting fakes, which is a race you cannot reliably win as the tools improve. The defense is to make a habit of verifying through a second channel, so that whether the voice is real or fake stops mattering. You will see how below.

The core idea: do not try to judge whether the voice or video is real. Assume you cannot tell, and verify the request another way instead. That single move makes the quality of the fake irrelevant.

Two · The four plays you will actually see

Learn the shapes, not the technology

Almost every AI-impersonation scam is one of four shapes. Learn them once and the specific technology stops mattering.

The loved one in trouble. A call, in a familiar voice, saying a family member has been in an accident, arrested, or kidnapped, and money is needed right now. The voice can be cloned from a few seconds of audio taken off social media.

The urgent boss. A call or video from someone who sounds or looks like your manager or a company executive, directing you to move money or share access immediately and quietly.

The famous endorsement. A video of a celebrity, official, or well-known investor promoting a platform or a can't-miss return. The face and voice are fabricated, and the offer underneath is the old guaranteed-returns lie.

The trusted authority. A voice or video posing as your bank, a government agency, or the police, using the borrowed authority to rush you into paying or handing over information.

Notice what every one of them needs to work: urgency, an emotional jolt, and a push to act before you check. Those three are the real fingerprints, and no amount of AI can hide them.

Grounding: the FTC warns that scammers use AI to clone a family member's voice from a short clip for family-emergency scams, and the FBI has warned of campaigns using synthetic audio to impersonate known figures. Verify any specific claim before repeating it.

Three · The one habit that beats all four

Verify on a channel they do not control

Here is the whole defense, and it is small enough to teach in a sentence: when a voice or a video asks you to move money, share access, or act urgently, do not respond on that same channel. Stop, and reach the real person or institution a second way that you choose.

If your grandson calls in a panic, hang up and call your grandson back on the number you already have for him, or call his mother. If your boss video-calls with an urgent wire, tell them you will confirm, then reach them on your normal work number or in person. If your bank calls, hang up and call the number on the back of your card. The scammer controls the channel they contacted you on. The instant you switch to a channel you control, their power is gone, because the real person will tell you the truth and the fake cannot follow you there.

The rule, in nine words: if it is urgent and about money, verify separately. Real emergencies survive a two-minute callback. Scams do not, because the moment you hang up to check, the pressure that made it work is broken.

The FTC's own advice against voice-clone scams is exactly this: do not trust the voice, and call the person back on a number you know is theirs, or reach them through someone else. It is not fancy. It is just reliable.

Grounding: FTC, "Fighting back against harmful voice cloning" and "Scammers use AI to enhance their family emergency schemes." The consistent advice is to verify through a known, independent contact and to treat demands for wire transfers, cryptocurrency, or gift cards as red flags.

Four · The family code word

A free tool that defeats a cloned voice instantly

There is one more thing, and it costs nothing. Agree on a code word with the people closest to you, a simple word or short phrase that only your real family knows, never posted anywhere. Then make a family rule: in any emergency call about money, anyone can be asked for the word, and a real person will know it.

A cloned voice can copy how your daughter sounds. It cannot know the word the two of you picked at the kitchen table. Asking for it on the spot ends the scam in one second, calmly, without an argument. Pair it with the callback habit and you have a defense that does not depend on your ability to detect a fake at all.

This is the part to bring to the people you love, especially older parents and grandparents, who are targeted most and often feel most ashamed afterward. Frame it gently: not "you might get fooled," but "let's set up a family word, so none of us ever has to wonder." Make it a shared habit, not a warning. The people who have a code word and a callback rule almost never lose the money, because the trap needs them to act before they check, and the rule makes checking automatic.

Five · The special case of the famous face

A video is not a fact

The endorsement play deserves its own note, because it feels different. A polished video of a trusted public figure promoting an investment can be completely fabricated now, voice, face, and all. Treat any investment pitch that arrives wearing a celebrity or official as false on sight, and judge the offer underneath by the same rule that has always worked: a promise of a guaranteed or unusually high return is fraud, no matter whose face is attached.

Verify the platform yourself in the regulator's own database, on EDGAR and Investor.gov, not on the glossy site the video sends you to. Regulators have warned specifically that fraudsters lean on AI buzzwords and fake endorsements to seem current and credible. The costume got better. The test did not change.

Grounding: SEC and Investor.gov, "Artificial Intelligence and Investment Fraud," on AI-themed pitches and fabricated credibility. Guaranteed high returns remain a top fraud hallmark.

If it already worked · The first hours

Move fast, and do not carry it alone

If a fake already got money from you, the steps are the same as for any scam, and speed helps. Contact your bank, card, or the payment app immediately and ask them to attempt a recall or chargeback. Save everything: the messages, numbers, and any recording. Report it free at reportfraud.ftc.gov and, if it was an investment, at sec.gov/tcr. And refuse the second wave: anyone who contacts you afterward promising to recover your money for a fee is the next scam, not a rescue.

Being fooled by a fake of someone you love is not a failure of intelligence. It is the result of a tool built to switch off your caution using the people you care about most. Tell one person you trust, and be gentle with yourself. Learning the callback rule and the code word, which you just did, is how you make sure it does not happen twice.

Keep this · The family safety card

Share these five with your household:

  1. Assume you cannot tell. A familiar voice or face is no longer proof. Do not try to judge the fake; verify the request.
  2. Switch channels. Hang up and reach the real person or institution a way you choose: a known number, in person, or through someone else.
  3. Set a code word. One private word only your real family knows. Ask for it on any emergency money call.
  4. Slow the urgency. Urgency plus money plus secrecy is the pattern. Real things survive a two-minute pause.
  5. Never pay a fake face. A celebrity or official in a video is not proof. Guaranteed returns are fraud regardless of who seems to say it.
You do not have to win a contest against the fakes. You just have to hang up and check. The voice can be copied. The callback and the code word cannot.

Educational only. This guide is general education, not legal, financial, or security advice. Tools and tactics change; treat specific examples as patterns at time of writing and verify current details with the official sources named. If you are in distress after a scam, please reach out to someone you trust. We hold no position and sell nothing here.

Real numbers. No hype. Receipts. · Alpha Vault · The Receipts Index · Carter Enterprise LLC · 2026