Start with the channel change
Say: “I do not approve payments or share codes during an unexpected call. I am going to hang up and verify another way.” Do not debate the caller’s identity.
Open the companion tool →Why this matters
Video quality, eye contact, and a familiar voice are presentation signals. They can be useful context, but they are not a secure channel. Asking someone to wave or turn their head may produce a clue; it should never be the only reason you send money.
A pause plan is deliberately boring. You state that you do not make high-stakes decisions inside an unexpected call, end it without negotiating, and call back through a number or app you had before the request. The separation is the protection.
The protocol
- Say: “I do not approve payments or share codes during an unexpected call. I am going to hang up and verify another way.” Do not debate the caller’s identity.
- Write down the request, the claimed reason, and any deadline while it is fresh. Do not use a number, link, or contact card supplied on the call.
- Call the person through a saved contact, start a new conversation from the official app, or ask a second trusted person to confirm independently.
- If the request concerns a bank or workplace, contact that institution using its published site or the number on an existing statement.
The bearing is a channel change
The app creates a plan, not a deepfake score. Its “ready” state means your alternate channel, pause sentence, and next human contact are written down — not that the video is genuine.
Limit first: This tool cannot inspect a live video or prove who is on a call. A convincing video can still be fraudulent, and an awkward video can still be real. If money or access is at stake, verify out of band.
Build your live call verification protocol →One human next step
For a suspected impersonation, preserve the message or call details and report it to the platform and relevant financial institution; if money moved, call the bank immediately.
Research log and safety checks
AI-assist path: not applicable to the core artifact. A model cannot safely turn a live-call appearance into proof; the concrete protection is a separate-channel protocol. §6.10: not a frontier-pace topic.
§0.5 protection result: category is deepfake-media-verification; crisis-adjacent: no; fear/urgency/scarcity toolkit used: no. The protective function is free and the paid feature is convenience only.