Why AI scams can be harder to recognise
Traditional advice often focused on poor spelling, strange formatting or obviously fake photographs. Those clues can still matter, but modern generative AI can produce polished writing, imitate familiar communication styles and create convincing synthetic audio or video.
The UK National Cyber Security Centre (NCSC) warns that scams are becoming smarter and that some can fool even experienced users. Ofcom has also highlighted the use of deepfakes in financial scams. The practical lesson is simple: judge the request and verify the identity, not just the quality of the content.
10 warning signs of an AI-assisted scam
1. An unexpected request for money
A relative, colleague, celebrity, company or supposed official suddenly asks you to send money. The request may arrive by text, email, social media, phone or video call and can sound extremely convincing.
Be particularly cautious if you are asked to use bank transfer, cryptocurrency, gift cards or another payment method that is difficult to reverse.
2. Artificial urgency
Scammers frequently create a deadline: a bill must be paid immediately, an account will be closed, an investment opportunity is about to disappear, or a family member supposedly needs emergency help.
Urgency is designed to prevent independent checking. A genuine organisation should tolerate reasonable verification.
3. You are told to keep the situation secret
Requests such as “don't tell anyone”, “don't contact the bank” or “keep this confidential until the transaction is complete” are serious warning signs. Fraudsters often try to isolate people from friends, relatives, bank staff or colleagues who might question the story.
4. A familiar voice is used as proof of identity
Voice cloning means hearing a recognisable voice is no longer sufficient proof that you are speaking to the real person. A supposed child, grandchild, manager or friend may appear to be calling from an unfamiliar number with an emergency.
End the call and contact the person using a number you already have. Families can also agree a private verification question or phrase in advance.
5. A realistic video or image is treated as unquestionable evidence
Deepfakes can imitate real people or make them appear to endorse investments, products or services. Visual oddities may sometimes expose a fake, but increasingly realistic content means the absence of obvious flaws proves very little.
Check whether the claim appears on the person's or organisation's genuine website and established accounts rather than trusting the video itself.
6. The contact method suddenly changes
A genuine conversation moves unexpectedly to WhatsApp, Telegram, another messaging app, a new email address or a new phone number. The person may claim their usual account is unavailable or compromised.
Changes of channel are not automatically fraudulent, but they should trigger independent verification before money or sensitive information is shared.
7. You are asked for passwords or security codes
Never provide a password, one-time passcode, PIN, recovery code or authentication approval simply because a caller or message appears legitimate. Criminals can combine convincing AI-generated communication with ordinary account-takeover techniques.
Contact the bank, company or service through its official app, website or a trusted number if you are unsure.
8. A payment destination changes unexpectedly
Be suspicious if a supplier, solicitor, tradesperson, colleague or family member suddenly gives you different bank details. Business-email compromise and impersonation scams often rely on replacing a legitimate payment destination.
Verify changed bank details using a previously established contact method before sending anything.
9. The offer seems unusually profitable or risk-free
AI-generated celebrity endorsements, fake news reports and cloned videos can be used to promote fraudulent investments, cryptocurrency schemes and bogus trading platforms. Claims of guaranteed profits, secret systems or unusually high returns deserve particular caution.
Do not assume an investment is genuine because a well-known person appears to recommend it.
10. The person discourages independent checking
This is one of the strongest warning signs. A fraudster may become impatient, aggressive or emotionally manipulative when you suggest hanging up, contacting the organisation directly or asking somebody else for advice.
A genuine request should survive independent verification.
The AimadeSafe STOP–CHECK–VERIFY method
1. STOP
Do not pay, click, disclose information or approve a login while you are under pressure.
2. CHECK
Ask what is unusual about the request: urgency, secrecy, changed details or an unexpected contact method.
3. VERIFY
Contact the person or organisation independently using details you already trust.
What not to rely on
There is no single visual or audio test that reliably identifies every AI-generated scam. Blinking, lip movement, distorted fingers, unnatural pauses and other anomalies can sometimes be clues, but genuine media can also contain glitches and synthetic media is improving quickly.
Likewise, an “AI-generated” label can be useful when present, but the absence of a label does not establish authenticity. Ofcom has examined approaches including AI labels, watermarking and provenance information, each with strengths and limitations.
If you receive a suspicious message or call
- Do not click links or download files from the suspicious contact.
- End the call or conversation if you feel pressured.
- Use the organisation's official app, website or a known telephone number to check the claim.
- For a supposed friend or relative, contact them separately using details you already have.
- Take screenshots or retain messages if doing so is safe and may help with reporting.
If you have already paid or shared information
Act promptly. If money or banking details are involved, contact your bank using its official contact details. If you disclosed a password, change it anywhere it has been reused and secure the affected account. If password reuse is a recurring problem, our Best Password Managers UK guide explains safer ways to manage unique logins. If you followed instructions to install software, follow current NCSC guidance and run appropriate security checks.
In England, Wales and Northern Ireland, fraud and cybercrime can be reported through Report Fraud. In Scotland, contact Police Scotland. Suspicious text messages can generally be forwarded free of charge to 7726.
Can security software stop AI scams?
No product can guarantee protection against every scam. Security tools may help identify malicious websites, suspicious messages, compromised credentials or other threats, but they should support — not replace — careful verification.
If you are considering additional protection, see our independent guide to scam-protection tools for UK consumers. We explain what the products can and cannot do and highlight renewal pricing as well as introductory offers.
Useful UK sources
- National Cyber Security Centre — how to spot scam messages and calls
- National Cyber Security Centre — what to do after sharing information
- Ofcom — deepfakes and methods used to identify AI-generated content
- Report Fraud — England, Wales and Northern Ireland
This guide provides general consumer information and is not financial, legal or cybersecurity advice. Scam techniques and official reporting arrangements can change, so check the linked official sources when taking action.