INTERACTIVE TOOL // TRUST NOTHING UNDER PRESSURE
Deepfake Scam Red Flag Checker
A familiar face is no longer proof. A familiar voice is no longer proof. Run a suspicious message through ten manipulation signals before fear, urgency or authority makes the decision for you.
This checker cannot authenticate a person, audio clip or video. It is a decision aid for spotting manipulation signals. If you are suspicious, break contact and verify independently using a number or channel you already trust.
Which red flags are present?
Your red-flag result
Tick every signal that appears in the call, message, voice note or video.
Why deepfakes change the old trust shortcut
People are trained from childhood to trust familiar voices and faces. Synthetic media attacks that shortcut. A scam does not need a perfect Hollywood-quality fake if the victim is already frightened, rushed or expecting bad news. A few seconds of convincing audio combined with the right personal detail can create the emotional certainty that makes verification feel unnecessary.
The safest response is therefore behavioural rather than forensic. Do not ask yourself only whether the voice “sounds real.” Ask whether the request contains scam mechanics: urgency, secrecy, unusual payment instructions, requests for credentials, changed bank details or resistance to independent verification. Those signals remain useful even as synthetic media gets harder to judge by appearance alone.
The family-emergency nightmare
Now place that problem inside our fictional AI-collapse scenario. Mobile networks are congested. A relative is away from home. A voice note arrives in their exact voice saying they are trapped and need money sent immediately to a new account. Seconds later a video appears to confirm it. The caller knows family details scraped from public posts. Meanwhile genuine emergency information is also circulating.
The defence is a rule agreed before the panic. Break contact. Call back using a number already saved or independently obtained. Use a family safe word or ask a question an impersonator is unlikely to answer from public information. Never treat caller ID, a familiar voice or an incoming video as sufficient authentication for a high-stakes request.
What to do if the checker shows high risk
Do not click links, install software, reveal codes or move money while the request is unverified. If the contact claims to represent a bank, government body, employer or other organisation, end the call and contact that organisation using its official website or a trusted number rather than information supplied by the caller. If banking details have already been shared or money moved, contact the bank immediately.
A low score is not proof that a message is genuine. Sophisticated fraud may avoid obvious pressure tactics. The result simply tells you how many recognised manipulation signals are present. For consequential requests — money, passwords, access, travel, evacuation, legal threats — independent verification should be normal even when the message appears calm and convincing.
Build the verification habit before the crisis
The easiest time to agree a family verification rule is before anyone is frightened. Choose a safe word or independent callback method, make sure older relatives understand it, and rehearse what happens if a message claims there is an emergency. The rule should be simple enough to remember when the caller sounds convincing and the story feels urgent.
Continue from here
Build this topic into a wider plan, then see how the same dependency could fail in an AI-collapse scenario.
UK scam-safety baseline
NCSC guidance highlights authority, urgency and emotion as common scam signals and recommends breaking contact and using an independent trusted channel when identity is in doubt.