What is a deepfake in fashion?
Realistic synthetic video, image or audio that depicts a person saying or doing something they did not.
In short
A deepfake is realistic synthetic video, image or audio that depicts a person saying or doing something they did not. For fashion companies it creates risks of fraud and reputational damage, and it raises consent questions when real people's likenesses are used in AI content.
How does it work in practice?
Deepfakes are made with generative AI models that learn a person's face, voice or movements from existing material and then produce new content. The same techniques that let brands create AI models or dub videos into new languages can be misused to impersonate real people.
Fashion-related risks include:
- A fake video of a designer or executive endorsing a product or making statements.
- Voice clones used to trick staff into urgent payments or data sharing.
- Fake ambassador or influencer content promoting counterfeits.
- Manipulated images of products or events that damage a brand.
Why does it matter?
Fashion relies heavily on faces, voices and personal brands. A convincing deepfake can spread quickly on social media before it is challenged. For B2B relationships, impersonation of executives or account managers can lead to fraudulent orders or payment redirection. Brands that use AI-generated people legitimately also need to avoid confusion with real individuals.
How does AI use it?
Generative AI produces deepfakes, but AI is also used to detect them by looking for visual or audio inconsistencies. Provenance tools such as content credentials and watermarking help show which media is authentic. The EU AI Act sets transparency obligations for deepfake content, requiring that such content is disclosed as artificially generated or manipulated.
Common pitfalls
Using a real person's likeness for AI-generated content requires their consent and clear contracts covering scope, duration and payment. Brands sometimes overlook this with models, ambassadors or employees. Other gaps include no verification process for unusual payment requests and no plan for responding quickly to fake content. Staff training, clear approval rules and monitoring of brand mentions reduce the risk.
Frequently asked questions
Is it legal to create an AI model based on a real person?
It generally requires the person's consent and a contract covering how their likeness is used. Rules differ by country, and transparency obligations may apply to the resulting content.
How can a fashion company protect itself from deepfake fraud?
It can verify unusual requests through a second channel, train staff to recognise impersonation and use provenance tools for official media. A rapid response plan helps limit damage if fake content appears.
