Scammers can now clone a loved one’s voice with AI to demand money

A phone rings, and the voice on the other end is unmistakable: a grandchild, a son or a daughter, frightened and pleading for money after an accident, an arrest or some other sudden crisis. For generations, con artists have exploited that scenario with the so-called grandparent scam, guessing at a relative’s identity and hoping panic overrides caution. Artificial intelligence has now removed the guesswork, giving fraudsters the ability to reproduce a real person’s voice from a few seconds of audio and use it to demand money in that person’s own tones.

Consumer-protection officials have warned that this new capability makes a long-running scam far more convincing, because the emotional shortcut people take when they recognize a loved one’s voice is exactly what the technology counterfeits. The result is a threat that turns publicly available audio — a social media clip, a voicemail greeting, a podcast appearance — into raw material for extortion.

How voice cloning changes the scam

Voice-cloning tools work by analyzing a short sample of a person’s speech and building a synthetic model that can then say anything the operator types. Only a small amount of audio is needed, and clips harvested from public videos or social accounts can be enough to produce a recognizable imitation. Because the underlying methods are related to the broader family of synthetic media known as deepfakes, the same advances that let software fabricate realistic video also let it fabricate realistic speech.

Armed with a cloned voice, a scammer can place a call that sounds like a specific relative in distress, often paired with a plausible story and a demand for immediate, hard-to-reverse payment. The cloned voice supplies the emotional authenticity that older versions of the scam lacked, making a victim far less likely to pause and question what they are hearing.

What regulators are warning

Consumer regulators have been explicit about the danger. The Federal Trade Commission has cautioned that scammers are using AI to enhance family-emergency schemes, noting that a caller who sounds exactly like a relative may in fact be a criminal wielding a cloned voice. The agency has urged the public to treat urgent money demands with skepticism no matter how familiar the voice, and it has taken steps to spur development of tools that can detect or deter voice-cloning abuse.

Those warnings are aimed especially at older adults, who are frequently the targets of family-emergency scams and who may be less aware that voice cloning is now cheap and widely available. The FTC has framed the problem as a case where technology has outrun most people’s mental model of what a phone call can prove.

The hallmarks of a cloned-voice call

Fraudulent calls tend to share recognizable features. The caller manufactures urgency, insisting that money must be sent immediately to handle an emergency, and often asks the victim not to tell other family members — a tactic meant to prevent the kind of quick check that would unravel the scheme. Requests for payment typically involve methods that are difficult to trace or recover, such as gift cards, wire transfers or cryptocurrency.

The distress in the voice, the pressure to act fast and the secrecy are all designed to keep a target from stopping to think. Recognizing that pattern is one of the most reliable ways to spot a scam even when the voice sounds genuine.

How to guard against it

Officials stress that the most effective defense is not technological but procedural: hang up and independently call the supposed relative back on a known phone number to confirm the story before sending anything. Many advisories also recommend agreeing on a private family code word or verification question in advance, so a genuine emergency can be authenticated in a way a stranger could not fake. The FTC’s guidance on fighting back against harmful voice cloning reinforces that verifying through a separate, trusted channel is the step that consistently stops these scams.

Reducing the audio available to would-be cloners also helps. Limiting public posting of voice recordings and tightening the privacy settings on social accounts shrinks the pool of samples a scammer can draw from, though it cannot eliminate the risk entirely for anyone whose voice appears in videos, interviews or voicemail greetings.

The technology behind these scams is advancing quickly, and detection tools that can flag synthetic audio remain imperfect and are not something an ordinary call recipient can rely on in the moment. That is precisely why officials keep returning to low-tech habits — calling back on a trusted number, agreeing on a code word, and refusing to act on urgent money demands without independent confirmation. As convincing synthetic voices become ordinary, the old assumption that a familiar voice on the phone is proof of identity no longer holds, and rebuilding a habit of verification is, for now, the surest protection.

This article was produced with the assistance of AI and reviewed by Morning Overview editors prior to publication.

More from Morning Overview