AI scams are becoming more convincing, and old warning signs are fading
Criminals are using artificial intelligence to make fraud look polished, persuasive and familiar. Fake retailers can now appear online with professional branding, convincing product pages and even a backstory that suggests a small family business selling quality UK-made goods. The result is a sharper form of consumer deception: buyers place an order, then either receive nothing at all or a cheap mass-produced item that bears little resemblance to what was advertised.
The shift matters because many of the instinctive checks people once relied on are less reliable. Misspellings, awkward phrasing and clumsy formatting used to be obvious signs of a scam. AI can now generate tidy copy, realistic images, and tailored messages at scale, making the fraud look far more legitimate at first glance.
Fake shops, cloned voices and personalised pressure
National Trading Standards has drawn up an AI scam watchlist for the first time, identifying several tactics now being used by fraudsters. One of the most visible is the rapid creation of cloned or entirely new websites that look slick and trustworthy with little cost or effort. These sites can mimic the appearance of genuine retailers closely enough to lure shoppers who are searching for bargain goods or local sellers.
Another tactic is the use of AI in romance scams. Here, the technology can be used to remember personal details, sustain long conversations over weeks or months, and make the interaction feel emotionally responsive. In fraud terms, the danger is not just volume but persistence. A system that remembers names, dates, family members and preferences can keep the deception alive long enough to lower suspicion and extract money or sensitive information.
One sentence captures the threat plainly: a convincing voice or face is no guarantee of identity.
Deepfake video and voice cloning also raise the stakes for family-based scams. A criminal can imitate a relative or close friend and claim to be in an emergency, asking for money with a tone that sounds urgent and familiar. AI-generated phishing emails, text messages and social media posts can be personalised on a massive scale, increasing the odds that at least one message will land with enough detail to trigger trust or panic.
Why the usual scam tells are no longer enough
Consumer group Which? warns that grammatical errors can no longer be treated as a dependable giveaway. That has practical consequences for anyone trying to decide whether a message, website or call is genuine. A polished scam no longer looks like a scam in the old sense; it can look like a brand, a business, or even a friend.
The mechanism is simple but powerful. AI reduces the time and skill needed to produce convincing fake material, allowing criminals to test many versions of the same scheme and tailor each one to the intended victim. In public health terms, this resembles an exposure that has become easier to spread and harder to detect. The harm is financial rather than infectious, but the logic is similar: wider reach, lower barriers, and faster transmission of risk.
Pressure is often part of the design. Criminals want people to act before they can think, compare details or contact someone else for a second opinion. Urgency can be built into a fake invoice, a social media plea, a romantic message or a voice note from a supposed relative in distress. Once panic takes over, scrutiny drops.
How to reduce the risk of being caught out
Trading standards advice focuses on slowing the decision down. Never send money, gift cards or personal or financial information to someone you have not met in person. If a purchase or request involves a shop, business or job advert, verify it independently rather than trusting the first link, message or account that appears online.
Checking whether a shop address is real can expose a fake operation quickly. For financial services, the Financial Conduct Authority’s Firm Checker can confirm whether a company is authorised to give financial advice. If a job advert looks suspicious, contact the business directly through a verified number or website rather than replying through the advert itself. Those steps sound basic, but they interrupt the fraud pathway before money leaves an account or private information is shared.
Trust should also be separated from sound and image. A voice on a call, a face on a video or a neatly produced clip does not prove identity. AI has made impersonation easier, so verification now has to rely more heavily on independent contact, known details and skepticism toward urgent requests.
Reporting fraud now matters more than ever
People who suspect fraud can report it through Report Fraud or call 0300 1232040. Citizens Advice Consumer Helpline can also provide guidance on 0808 223 1133. These routes matter because large-scale AI-enabled scams depend on repetition, and reporting helps authorities map the patterns that criminals are using.
The wider lesson is straightforward. Online fraud no longer depends on crude bait or obvious errors. It can arrive as a clean website, a caring voice, a convincing video or a message that feels exactly like the real thing. The safest response is still the slowest one: pause, verify and refuse urgency until identity and intent are confirmed.