Responsible use of AI in advertising
- Fernanda Tissot
- Jan 15
- 4 min read
It is well known that generative Artificial Intelligence (AI) has been gaining increasingly more ground across all business sectors, particularly in advertising.
This is because using generative AI platforms such as DALL-E, Sora, Midjourney, among others, can significantly reduce the time and financial resources spent on advertising creations.
On the other hand, the indiscriminate use of these platforms can introduce significant legal and reputational risks that must be taken into account when producing ad creative that uses generative AI. To illustrate these risks, we can cite the renowned Brazilian agency DM9, which lost the award it won in the "Creative" category at Cannes Lions after it was discovered that AI-generated and manipulated content was used to simulate real events in the campaign[1].
In the field of intellectual property, there are numerous discussions surrounding generative AI—whether regarding ownership of rights (who owns the rights to an AI-generated "work"?), authorship (who is the author of an AI-generated "work"? The individual who entered the prompts/commands or the "machine"?), or potential copyright infringement and copying of material developed by the system.
At this point, significant uncertainty surrounds the application of copyright to AI-generated content, as there is no consensus on whether such content is copyright-protected, given that Brazilian Copyright Law considers only human beings as authors.
Thus, the existence of authorship will always depend on the degree of human intervention in the creation (if AI is used merely to assist the author and is not decisive in generating the content, copyright protection can be justified). Furthermore, if it is determined that the content is not protected by copyright, it will enjoy no protection at all, allowing third parties (including competitors) to use identical or "inspired" campaigns, once again creating reputational risks for companies.
Therefore, beyond intellectual property, concerns about the risks involving the use of AI are growing in Brazil and are at the center of legislative discussions. Bill No. 2,338/2023[2], authored by Senator Rodrigo Pacheco (PSD/MG), is currently under debate; as stated in its Article 1, it seeks to protect fundamental rights, foster responsible innovation and competitiveness, and ensure the implementation of safe and reliable systems.
To achieve this goal, the Bill sets forth numerous principles, definitions, and risk categorizations regarding the use of generative AI that will be instrumental in guiding AI regulation in Brazil, following international trends on the subject.
While legislative, academic, and judicial discussions remain ongoing, it is important to highlight key precautions that should be taken to mitigate exposure to legal risks.
In this regard, particularly for advertising campaigns, we recommend several best practices, including:
Use of prompts
Avoid using prompts (commands given to generative AI) that could potentially infringe upon intellectual property or image rights.
Examples: Prompts such as "generate a song in the style of the band KISS"; "generate an image of a person who looks like Chico Buarque"; "generate a video in the style of the series The Last of Us", etc.
Document the prompts used
When developing a campaign using generative AI, it is important to document the prompts used across all generative AI platforms. Should a legal defense be necessary in a lawsuit, this allows a company to demonstrate good faith and, if statutory requirements are met, potentially establish authorship.
Since there is no consensus on authorship and ownership, it is recommended to log every step of both human and non-human intervention via notarial certificates (atas notariais) or digital evidence registration tools.
Sensitive information
Avoid inputting sensitive data into generative AI platforms, such as:
– Personal data (tax ID numbers [CPF], full names, etc.)
– Trade secrets (internal and confidential company data)
– Images and videos of individuals without proper authorization
Such data may be used by the AI model for training/learning purposes and could appear in responses provided to third parties.
Analysis of terms and conditions of generative AI platforms
Platform terms and conditions (the documents setting forth rules of use) often prohibit commercial use of AI-generated outputs. Therefore, it is necessary to thoroughly evaluate these terms of service to understand all applicable rules and usage limitations.
Attention to contractual clauses with advertising agencies
If the campaign was created by an external agency, it is highly recommended to include contractual provisions requiring documentation of the prompts used or even prohibiting the use of AI altogether.
Additionally, we recommend including indemnity clauses seeking indemnification for damages caused to your company in the event of third-party rights violations committed by the advertising agency.
Finally, it is important to note that standard contractual clauses transferring financial copyright/economic rights (cessão de direitos patrimoniais de autor) for advertising campaigns—common in such transactions—may not apply, as there may be no copyrights to transfer in the first place.
Development of internal best practices manuals for generative AI use
It is always recommended to establish an internal manual of best practices for generative AI use within the company. This manual will outline recommended uses, highlight associated risks, and serve to demonstrate good faith in the event of judicial disputes.
Have questions on how to use generative AI responsibly and with lower risk? Speak with the team at Ritter Advogados.
Comments