Skip to content
ReputationKiln

A reference on online reputation  ·  Calm, sourced, free


A deepfake of you or your business

By ReputationKiln Editorial · Published

Generative AI can now clone an owner's face and voice from a few seconds of footage, and the fakes are good enough that you usually cannot tell by watching. They get used three ways: to defraud your own staff by faking your voice or face on a "transfer this money" instruction, to scam your customers with a deepfaked "endorsement" of some investment, and to damage your reputation with a fabricated clip of you saying something you never said. The losses are large and climbing fast.

Because you cannot reliably detect a deepfake, the defence is the one the whole trust economy is moving towards: verify through known channels, prove your own genuine media, tell your customers your real channels, and report fast.

What to do

  • For a "the boss wants a payment" deepfake: always verify out of band. Hang up and call back on a known number, confirm any money or password request through a separate channel, and never act on urgency, because urgency is the tell. Agree a verification word for payment instructions in advance, and never authorise a transfer on a voice or video call alone.
  • For a deepfaked endorsement using your face: your control over your own name, voice and likeness is the main legal lever, alongside reporting it under the platforms' manipulated-media and impersonation rules. Get ahead of it by telling your customers plainly what you will never do, for example that you will never ask them to invest in anything.
  • For a reputation-damaging clip: the defamation and recourse routes apply, and the deepfake itself is evidence.
  • Report to the relevant fraud authority and the police.

The limit, and the durable defence

Be careful what you claim about the law, because it is a fast-moving patchwork. A recent United States law forces fast removal of intimate deepfakes specifically, and it does not cover general business or endorsement fakes, so do not assume it does. The honest, durable defence is the same proof-not-trust move as everywhere here: you cannot stop a deepfake being made, and you usually cannot detect one, so you win by being verifiably real, by signing and owning your genuine media, and by making sure your people and customers verify before they act.

Sources

  1. A finance worker was deceived into paying about USD 25.6 million after a video call in which other participants were deepfake recreations of colleagues. — CNN Business, Arup revealed as victim of deepfake scam (2024). https://www.cnn.com/2024/05/16/tech/arup-deepfake-scam-loss-hong-kong-intl-hnk · checked 2026-06-04
  2. Because detection cannot reliably keep up, the durable defence is provenance: signing genuine content so it can be verified as real. — Coalition for Content Provenance and Authenticity (C2PA). https://c2pa.org/ · checked 2026-06-04