AI trends & brands: 5 legal risks you should know

When Internet users take ownership of a brand's universe thanks to generative AI

The phenomenon is recent but massive. Influencers use generative AI tools to create visuals stamped with the colors of a brand -packaging, color codes, editorial atmosphere - then broadcast them on social networks. All without a mandate, without a contract, without authorization.

The most visible example to date is the “Rhode Girls” trend, in which Internet users reproduced the visual identity of Rhode Skin, Hailey Bieber's cosmetics brand. But the mechanism can be replicated to any brand with an identifiable graphic universe.

The phenomenon is attractive from a marketing point of view. In law, it raises at least five major points of attention.

1. Reproducing the visual identity of a brand without authorization remains a potential infringement

Using the visual identity of a brand - name, logo, color codes, graphic universe - without authorization can constitute an infringement of the owner's intellectual property rights, whether it is trademark infringement or parasitism.

In the case of the Rhode Girls trend, the visuals generated were widely perceived as flattering, which probably explains the lack of reaction from the brand. But this tolerance is by no means a legal blank check: it remains a decision of opportunity, revocable at any time.

2. The target brand cannot freely reuse content generated by Internet users

If a brand wants to exploit content generated spontaneously by Internet users to promote its own products, it cannot do so freely. The use of these visuals for commercial purposes implies the authorization of their authors and, most likely, remuneration.

This is a point that marketing departments and agencies must anticipate: surfing on a viral trend does not exempt from securing the rights to the content used.

3. AI visuals can damage brand image

Not all AI-generated content is flattering. If certain visuals associate a brand with inappropriate staging, controversial personalities or trivialize luxury positioning, the trademark owner can take legal action against broadcasters.

The ease of creation offered by generative AI mechanically multiplies this risk: anyone can now associate a brand with a context that it has never validated, in a few seconds and on a large scale.

4. The line between viral trends and disguised advertising is increasingly blurred.

In France, the law of June 9, 2023 governing commercial influence requires that all promotional content shared online be clearly identified as such.

However, these AI trends create a gray zone: the content looks like advertising, carries the codes, but is not (officially) sponsored by the brand. The question would be particularly acute if a brand was at the origin of such a trend, for example by broadcasting an “advertising prompt” intended to be taken up massively by Internet users.

5. AI tools that replicate brand worlds raise the question of training

For the Rhode Girls trend, several generative AI tools - including GlamApp and MIUIA - made it possible to produce the visuals. These tools are clearly capable of accurately reproducing the visual identity of protected brands.

This raises the now recurring question of respecting intellectual property rights in the training of generative AI models. If a tool is able to accurately reproduce the graphic universe of a brand, it is because its training data contained protected elements - and the question of the authorization of the rights holder remains unanswered.

Not all brands will let it happen

For the time being, Rhode Skin has let the trend develop without litigation. Not every brand will make that choice. A luxury brand concerned with maintaining the exclusivity of its image, or a brand operating in a regulated sector where the distribution of uncontrolled content can lead to specific offences, will have every reason to act.

Brands, agencies, and content creators have an interest in anticipating these risks rather than discovering them at the litigation stage.