AIMS project
AI Incidents in Marketing
AI Incident Database // Deepfakes & Unauthorised Likeness

The Taylor Swift deepfake and Le Creuset's brand safety problem

When scammers used an AI-generated version of Taylor Swift to promote a fake Le Creuset giveaway, the brand became part of a fraudulent campaign it had neither created nor authorised.

Incident typeDeepfake, unauthorised likeness, fraudulent advertising
DateJanuary 2024
RiskBrand reputation, fraud, consumer trust, unauthorised brand use

In January 2024, fake social media advertisements appeared to show Taylor Swift promoting a giveaway for Le Creuset cookware. The ads used manipulated video and an AI-generated version of Swift's voice to tell viewers that thousands of cookware sets were being given away because of a packaging error.

The promotion was not genuine. Le Creuset stated that it had no partnership with Swift for the giveaway and advised consumers to rely on the company's official channels for genuine promotions. Some versions of the scam directed users to websites that requested personal information and a small payment described as a shipping fee.

Le Creuset was not responsible for producing the content, but its brand was still drawn into the incident. Scammers used the credibility of both a well-known celebrity and a premium consumer brand to make the offer appear plausible, creating a brand-safety problem that originated entirely outside the company's own marketing operations.

What went wrong

This case illustrates a different category of AI marketing ris: third parties using generative AI to imitate a celebrity and falsely associate that likeness with a legitimate brand.

Traditional marketing controls are usually designed around content that an organisation creates, approves or purchases itself. Deepfake scams bypass these controls and can create a real reputational issue for the company.

For marketing and brand teams, this broadens the scope of brand safety. Governance cannot focus only on reviewing authorised campaigns. Organisations also need monitoring, escalation, and response processes for content that misuses their name, products, logos, or partners without permission.

Governance questions

  1. Does your organisation monitor for AI-generated content that uses its brand name, products, logos, influencers, or celebrity partners without authorisation?
  2. If a fraudulent deepfake campaign appeared using your brand tomorrow, who would verify the incident and coordinate the public response?
  3. Do your contracts with influencers, ambassadors, and celebrity partners address unauthorised synthetic uses of their likeness or voice and define how the parties should coordinate if such an incident occurs?

Learning outcomes

After discussing this case, participants should be able to:

  1. Explain the difference between risks created by a brand's own use of AI and risks created by third-party AI-generated content.
  2. Identify how deepfake advertising can exploit the combined credibility of a public figure and a legitimate brand.
  3. Describe the elements of a rapid-response process for fraudulent AI-generated advertising involving a brand.
  4. Explain why brand monitoring may need to cover unauthorised synthetic content as well as authorised advertising.
  5. Assess whether existing influencer, ambassador, and celebrity-partnership governance covers deepfake and synthetic-media incidents.

Discussion questions

  1. What responsibility does a brand have to respond when it is itself a victim of a deepfake scam?
  2. Which signals should trigger escalation: use of the brand name, paid distribution, requests for payment, customer complaints, or evidence that people have already been deceived?
  3. How should responsibility for detection and response be divided between marketing, legal, cybersecurity, customer service, and platform partners?