The short answer
Convincing fake voices, faces and text are now cheap and quick to produce. The old advice about spotting bad spelling or odd hands no longer works reliably. What still works is refusing to verify someone through the channel they contacted you on, and checking anything urgent through a route you chose yourself.
The old tells have gone
Poor spelling, strange grammar, six-fingered hands, robotic voices: all of these were genuine signals a few years ago. They are not any more, and advice still built on them gives people false confidence.
A voice can be cloned from a short clip of someone speaking. Video calls can be faked in real time. A scam message can be written in fluent, friendly, regionally accurate English for nothing.
So the question has changed. It is no longer “does this look fake?” but “how do I know who I am actually talking to?”, and those need completely different answers.
BEEP / THE MYTH-BUSTER
I am the reason your usual checks stopped working. The one thing I cannot fake is being reachable on a number you looked up yourself. Everything else is now cheap.
The one check that still works
Never verify someone using the contact details they gave you. Not the number in the message, not the link in the email, not the person on the call you are already on.
Hang up and call back on a number you find independently — on your card, on the official website you typed in yourself, on a previous statement. If they are genuine, they will expect this and be untroubled by it.
For family, agree a word or a question in advance that a stranger could not answer. It sounds theatrical until the day somebody rings in your daughter's voice asking for money.
What the common approaches have in common
- Urgency
- Act now, or something bad happens. Urgency exists to stop you checking.
- A new payment route
- New account details, a different card, an unusual method. Always a warning.
- Secrecy
- Do not tell anyone, do not tell your bank. Genuine organisations never ask this.
- A familiar voice
- Cloned from a few seconds of audio. Recognition is no longer proof.
The shape of the pressure rather than a checklist to tick. Any one of these is enough reason to stop and verify separately.
Checking an image or video
Provenance beats forensics. Where did this come from, who first posted it, and does any organisation with a reputation to lose also report it? That is a better test than staring at the pixels.
Reverse image search still helps, particularly for old pictures recycled as new events. It is free and takes seconds.
Some content now carries built-in provenance data showing how it was made, and some platforms label it. Useful when present. Its absence proves nothing at all.
What to actually do
- Stop when you feel urgency. That feeling is the product being sold to you.
- Verify through a channel you chose, never one you were given.
- Agree a family phrase now, before you need it.
- For news, ask who first published it rather than whether it looks real.
See it for yourself
Reading about it only gets you so far. This one is written to show you the thing described above, rather than describe it again.
Here is a message I have been sent. Point out anything in it designed to make me act quickly, and tell me what I should check before I reply to any of it.
Then paste it into whichever assistant you use — ChatGPT, Claude, Gemini or Copilot. Not sure which? Meet them all.
If you remember nothing else
- Spelling, hands and robotic voices are no longer reliable tells.
- A voice can be cloned from seconds of audio, so recognition is not proof.
- Never verify someone through the details they supplied.
- Urgency, secrecy and a new payment route are the constants across scams.
- For images, provenance is a better test than inspecting the picture.
Where this came from
- National Cyber Security Centre, phishing and scam guidancechecked 26 September 2026
Anything here that turns out to be wrong is worth telling us about. Figures that move — prices, sizes, energy use — are kept in the registry with their dates, rather than written into the prose where they would quietly go stale.
Questions about “Scams and deepfakes”Asked by readers, answered by us. Open to read them.
Open to load them.
Ask about this page and we will answer it here, where the next person can read it too.
Nobody else can answer you, and you cannot answer anybody else. This is not a forum — we would rather say that than run a room we cannot watch at three in the morning.
Your question goes up with the answer and without your name. Nothing identifies you, ever.
If we cannot answer it we will tell you why, and that one stays between us.