Deepfake
Also called deep fake, deepfake video
What is a deepfake?
A deepfake is a video, audio clip or image that has been created or altered with AI to realistically show a real person saying or doing something they never did. Common forms are face swaps, lip-synced fake speeches and cloned voices. The word combines 'deep learning' and 'fake' and dates from around 2017.
Deepfakes use the same technology as legitimate synthetic media: face generation, lip sync and voice cloning. What makes something a deepfake is that it depicts a real, identifiable person in a way that is deceptive or made without their consent.
Deepfakes are used in scams (a fake executive voice asking for a payment), fake celebrity endorsements in ads, harassment and political misinformation. Many places now have laws on non-consensual deepfakes and AI in political or commercial content, and major platforms require labels on realistic AI media.
Why a deepfake matters
For businesses, deepfakes are a risk in two directions. Scammers may impersonate your brand or staff, and fake endorsements using famous faces are a common ad scam that platforms remove and regulators pursue.
If you use AI presenters or voices in your own content, keep clear of anything that could be mistaken for a real person who did not agree to it. Consent, disclosure and content credentials protect your reputation.
How to spot a deepfake
- Check the source: who posted it first, and does the real person or brand confirm it?
- Look for odd blinking, blurry edges around the face, mismatched lighting or teeth.
- Listen for flat emotion, odd breathing or words that do not match mouth shapes.
- Look for content credentials or platform AI labels.
- Be most skeptical of urgent requests for money or data, however real the voice sounds.
Deepfake FAQ
Are deepfakes illegal?
It depends on the content and where you are. Many jurisdictions ban non-consensual intimate deepfakes, and fraud, defamation and false advertising laws apply to deepfakes too. Laws in this area are changing quickly.
What is the difference between a deepfake and synthetic media?
Synthetic media is any AI-generated media, used openly or with consent. A deepfake is synthetic media that realistically impersonates a real person, usually to deceive.
Related terms
Synthetic media is any image, video, audio or text created or substantially changed by AI, from AI voiceovers to generated photos and avatars.
Voice cloning is AI that copies a specific person's voice from a recording sample so it can speak any new text in that voice.
An AI avatar is a computer-generated person or character that can speak a script on video, with lip movements synced to an AI or recorded voice.
Content credentials are tamper-evident metadata, based on the C2PA standard, that record who made a file, with which tools and whether AI was used.
An AI watermark is a visible mark or hidden signal embedded in AI-generated content so software can later identify it as AI-made.
Brand safety is protecting a brand's reputation by keeping its ads and content away from harmful contexts and its own content free of risky claims.