骗子是怎么把 AI 用在诈骗和欺诈上的？How Criminals Misuse AI for Scams and Fraud

Cloned voices staging a "kidnapping," phishing written to fit you alone: cybercrime has been handed a significant upgrade. What follows is a walk through the exact tactics aimed at your money and your identity, plus the measures that genuinely shut them down.

You reckon a con would never fool you? In 2026, that confidence may be misplaced. The sloppy emails full of typos and the stilted robocalls have largely been abandoned. What criminals reach for now is AI: voices duplicated, phishing copy written without a single flaw, and relationships patiently cultivated with a target over weeks or months until they feel entirely genuine.

Tracking the ways these tools get bent to criminal use is what DSH Plugin Hub does, so the lesson does not have to be learned the painful way. Condensed: no new species of fraud has appeared — the existing ones simply became cheaper to operate, tougher to detect and unnervingly believable.

01How AI Scams Changed: What Makes 2026 Its Own Thing

Volume was the old strategy — fire thousands of interchangeable messages into the void and count on a tiny fraction taking the bait. AI inverts that. The net is no longer wide and crude; it is tight, tailored and quietly alarming, stitched together from genuine facts about you that were harvested from whatever you had already posted publicly.

Fraud powered by AI is one piece of a larger picture. Privacy, bias, data security — the whole terrain of digital risk gets a fuller treatment in our guide to everyday AI risks.

Official statistics now capture the scale. When the FBI's Internet Crime Complaint Center (IC3) published its 2025 annual report, AI appeared as a standalone fraud category for the very first time — and the figure is broadly regarded as an undercount, because in most cases victims never learn that AI played a part. The full FBI announcement has the details.

Source: the 2025 Internet Crime Report, published by the FBI's Internet Crime Complaint Center (IC3). Separately, Deloitte's Center for Financial Services expects these tools to grow cheaper and simpler to operate — and US losses from AI-enabled fraud to rise toward $40 billion annually by 2027.

02Cloned Voices and the Fake "Kidnapping" Call

Imagine the phone buzzing. On the line, sobbing, is your child or your partner, telling you there has been an accident or that they have been seized. The voice is unmistakable — the pitch, the rhythm, the particular way they pronounce your name. Only none of it is happening. A machine is producing it live.

Inside a Fake-Kidnapping Call, Step by Step

1. The opening: an unfamiliar number lights up the screen, and then a voice you know, in tears — "Mom, please, they've got me!"
2. The squeeze: a second voice joins in, the supposed "kidnapper," insisting on an immediate wire transfer or gift cards as the price of your family member's safety.
3. The isolation: hanging up is forbidden, calling anyone else is forbidden — and panic does the rest, crowding out clear thought.

03Phishing Written for One Person Only

The typo-riddled "Nigerian Prince" message is a relic. Spear phishing, as the AI-assisted variety is known, puts language models to work combing your LinkedIn profile, half-forgotten posts and public records — after which a message drafted for you and nobody else lands in your inbox.

Businesses face this too, not in theory but in the numbers: FBI data ties more than $632 million of 2025 investment-fraud losses, along with roughly $30 million lost to business email compromise, straight to tactics that used AI — and researchers in security expect both totals to keep rising. To report an attempt yourself, the official US route is the FTC's ReportFraud.ftc.gov.

04Romance Fraud and Digital Catfishing

Cruelty was always part of romance fraud; what AI added is scale. Nobody has to spend hours charming one victim after another by hand any more. A chatbot tuned to sound emotionally present keeps dozens of "relationships" running in parallel — birthdays recalled, small questions about your day asked, trust assembled so gradually that defences come down on their own.

When the subject of money finally surfaces — dressed up as a medical crisis, or a plane ticket for the long-awaited first meeting — it no longer reads as fraud. It reads as coming to the aid of somebody you love. Our starter guides lay out the safety habits worth having first.

05Bogus Investments and Deepfake Crypto Schemes

Financial crime is being reshaped as well. With deepfake video, a convincing "live" investment seminar can be staged around a duplicated billionaire or finance celebrity — and the quality is no longer the crude cut-and-paste it once was.

06Recognising an AI Scam in 2026

With scams this well finished, instinct is no longer a shield, and hunting for spelling mistakes gets you nowhere. The habit that does work: assume any urgent request you did not expect is suspect until proven otherwise. Some signals worth knowing:

• Oddities in the audio: breaths tend to be missing where they ought to fall, a strange hiss sits in the background, and interruptions provoke odd responses.
• Oddities in the video: on a deepfake call, blink patterns look wrong, the outline of the face softens into blur, or the lips slide a fraction behind the sound.
• Pressure plus secrecy: an instruction to move at once and inform nobody is, by itself, one of the loudest warning signs there is.
• Payment methods that do not fit: a genuine company or government body has no reason to demand cryptocurrency, a wire transfer or gift cards.

07A Protection Routine That Works

None of that leaves you defenceless. A handful of habits is enough to make you considerably harder to hook.

If you or somebody close to you has already been approached, the FTC Consumer Alerts page carries up-to-date guidance written in ordinary language, and a report can be filed straight away at ReportFraud.ftc.gov.

08Questions We Get Asked