How to Verify Someone’s Identity When Voice and Video Can Be Faked

Reduce what an impersonator can use

AI makes stolen identity information more persuasive, not magically more authentic. The practical goal is to reduce the data available to an impersonator and make important decisions depend on verification rather than familiarity.

Synthetic voice and video can imitate familiar people well enough to support impersonation scams. Keep that in mind when deciding what to protect first: the aim is to make stolen information less useful and important requests harder to fake successfully.

What matters in practice

Synthetic voice and video can imitate familiar people well enough to support impersonation scams.

Public information becomes more useful to a scammer when it can be assembled quickly into a believable story. Reducing unnecessary exposure removes details that make an impersonation feel convincing.

Independent verification works because it forces the claimant to succeed through a channel the attacker may not control.

Your primary email and recovery routes deserve special protection because they can unlock many other accounts. Strong authentication and well-protected recovery methods reduce the damage from a single stolen password.

Payment-detail changes and urgent secrecy requests are especially important moments to slow down and verify.

Identity checks need a fallback that does not depend on the suspicious message itself. Known numbers, agreed phrases and separate account channels give you a way to verify a claim when voice or video cannot be trusted on appearance alone.

Where impersonation risk really comes from

Identity risk sits at the intersection of exposed information, account security and human trust. A single privacy measure cannot solve all three, so claims of complete protection deserve particular caution.

• No single family code word solves every impersonation risk, especially if it is shared carelessly.
• Visual artefacts and robotic audio are becoming less reliable as detection cues improve.

What could change the answer?

Revisit the assessment after an account breach, a change in recovery details or a large increase in public exposure. Identity protection is a maintenance job, not a one-time privacy clean-up.

What you can do now

Protect the accounts and information that make impersonation useful. Then add a separate verification habit for requests involving money, access or sensitive information.

1. For family emergencies, call the person back using a saved number.
2. For organisations, verify payment or account changes through an established second-person process.
3. Treat urgency, secrecy and changed contact details as reasons to add verification, not skip it.

A four-question reality check

When a claim sounds dramatic, work through these four questions before deciding how much weight to give it.

Bottom line

You cannot make your identity impossible to imitate, but you can make impersonation less convincing and much harder to turn into account access, money or sensitive information.

In the AI era, “I saw their face” and “I heard their voice” are weaker evidence than they used to be. For consequential requests, identity should be verified through a separate trusted channel, known contact route, shared secret or established organisational process.