Legal Risks of Deepfake and Impersonation Advertising, and the Initial Response for PR Legal

Hello, I'm Noriaki Asato, Representative Attorney at LegalAgent.

With the spread of generative AI, content that imitates a person's face, voice and manner of speaking can now be created easily. As advertisements and social media posts impersonating celebrities, executives, doctors and attorneys have become a problem, startups too increasingly need to be aware of the risks in PR, recruitment and social media operations. This is not only about preventing harm from malicious advertising; it is also a question of checking, when the company is the one creating materials, how far it can use other people's likenesses, names and titles.

Risks of Using Other People's Faces, Voices and Names in Advertising

Using another person's face, voice or title in advertising often requires that person's consent. Representations that meet the requirements of the Premiums and Representations Act are a problem, such as a business that supplies goods or services pretending to have a celebrity's endorsement and thereby showing its content to be significantly better than it actually is. In addition, the use of a person's appearance involves personal interests in one's likeness, and acts that exploit the customer-attracting power of a name or likeness involve the right of publicity. The use of a voice cannot necessarily be treated in the same way as the use of a likeness, and the impact on personal interests and other rights should be examined according to the specific manner of use. Defamation or damage to credibility, and infringement of copyright or trademarks, may also arise. Furthermore, falsely using a company's name or logo can be subject to injunctions and claims for damages under the Unfair Competition Prevention Act, as causing confusion with a well-known indication or misappropriating a famous indication.

Even with an AI-generated person, there is no room for complacency if it evokes a real person. In particular, advertisements made to resemble doctors, celebrities or investors may lead consumers to mistakenly believe that the person has endorsed or is involved with the product. For expressions that borrow the credibility of a third party, simply indicating that the content is AI-generated does not immediately make it lawful, so check in advance the person's consent, the scope of the contract and the content of the representation.

Initial Response to Impersonation and Preservation of Evidence

When the company or its representative falls victim to an impersonation advertisement or post, proceed with preserving evidence and containing the spread of harm. Save the URL, screenshots of the screen, the ad ID, the date and time of display, and the destination it leads to. This is because if the content is deleted without evidence being kept, it becomes difficult to trace later. Promptly secure the necessary evidence, and proceed with takedown requests to the platform, warnings, and identifying the advertiser. Consulting the police and posting warnings on the official website should also be considered in parallel, depending on the urgency. When relying on copyright or trademarks, proceed along the appropriate route after confirming the company's actual use and the rights involved.

Contract Management with Advertising Agencies and Production Companies

The risks of deepfake and impersonation advertising should also be managed in contracts with advertising agencies and production companies. Set out the types and terms of the AI tools used, whether the generated output infringes any rights, the scope of use of materials, liability and indemnification in the event of a breach, and emergency takedown procedures. In fields such as finance, healthcare and investment in particular, advertisements using authoritative figures have a large influence on consumers, so careful arrangements are needed.

Usage Rules for Internal PR and Recruitment PR

In internal PR and recruitment PR as well, there are situations where AI materials are used, such as employee-like images, synthesized voices of the representative, or event-like images. Making these look like real people or real events can cause misunderstanding. Decide in advance whether to indicate that they are AI materials or to use expressions that will not be mistaken for real people. In PR legal work, it is important to assess not only copyright but also whether the communication avoids damaging the company's credibility.