What Are the Risks of Using AI to Generate Advertisements?
With the development of AI technology and related industries, its high efficiency and low cost have won the favor of many enterprises. Using AI to generate advertisements has become increasingly common. However, if enterprises ignore the compliance of the content of advertising promotions, they may encounter the legal risks such as administrative penalties, and damages to their brand and reputation. This article summarizes the risks of using AI to generate advertisements and the corresponding countermeasures.
Risk 1: It is Difficult for enterprises to protect AI-Generated Advertisements on the ground of copyright.
Compared with advertisements created by humans, AI-generated advertisements are at a higher risk of failing to meet the originality requirement for works under the “Copyright Law”. Although advertisements created by humans may also draw on works they have seen or heard before, just as there are no two identical leaves in the world, works created by humans mostly have different degrees of originality unless they are plagiarized or imitated. The current mainstream judicial view on AI-generated works also depends on their originality. That is to say, in theory, AI can be regarded as a tool similar to a ruler or an electronic drawing tool. If the instructions given to AI, including prompts, parameters, etc., are selected and arranged with originality, the AI-generated work shall be a work protected by the “Copyright Law”. However, due to the large number of reference elements in the AI database, if the prompts, parameters, etc., lack originality and are merely a simple combination of elements, such works will usually not be recognized as works in the sense of the “Copyright Law” and thus cannot be protected on the ground of copyright. In addition, it is also important to prove the process of originality in selecting and arranging elements such as prompts and parameters. For example, in the case of (2024) Su 0582 Min Chu No. 9015, the court held that: “The plaintiff failed to provide the original records of the creation process and could not prove that it had made sufficient personalized choices and substantive contributions to the AI-generated results, so the involved pictures do not constitute works in the sense of the Copyright Law.”
Risk 2: Advertising infringement may lead to liability for compensation.
Since AI is not a simple electronic drawing tool, its intelligent part lies in the fact that it is fed a large amount of data (graphic, text, and audio elements). Therefore, if the selection or arrangement of prompts, parameters, etc., lacks originality, that is, there is an unauthorized “image embedding” or “audio embedding”, in addition to the aforementioned Risk 1, it may also be suspected of infringing the rights of others.
Risk 3: Violating software usage agreements or platform agreements may lead to liability for breach of contract.
When using AI software, enterprises usually have to sign agreements with software rights holders, such as user service agreements or user license agreements. If the software is used beyond the scope authorized or permitted by the agreement, for example, the agreement stipulates non-commercial use but the enterprise uses the AI software to generate commercial advertisements, it shall be deemed a breach of contract. In addition, some enterprises may publish advertisements on large platforms. If they violate the platform agreements, such as failing to meet the platform’s special requirements for AI-generated advertisements, it may also constitute a breach of contract.
Risk 4: The content or the label of the AI-generated advertisements does not comply with relevant laws and regulations.
AI software cannot automatically identify whether the content complies with the relevant requirements of the “Advertising Law”, such as the authenticity of the stated content, ethics and morality, extreme words and etc. Therefore, if manual review is not conducted, there is a risk of non-compliance. In addition, the “Measures for the Labeling of AI-Generated Synthetic Content” came into force on September 1, 2025. According to this regulation, enterprises shall add explicit or implicit labels to the text, audio, images, videos, and other content generated by AI. This is also a point that enterprises tend to ignore or are unwilling to label, thus triggering risks.
To sum up, if an enterprise intends to use AI to generate advertisements, it is recommended to take the following four compliance review steps.
- To review the agreements with advertisers, AI software holders, and advertising platforms, pay attention to intellectual property clauses, permission for commercial use, allocation of liability for infringement, etc.
- Pay attention to work traceability. Enterprises should keep records of the process of using AI software to generate works, such as the initial copy, ideas, prompts, embedded images, and records of multiple parameter adjustments, modifications, optimizations, screenings, and etc. The purpose is to demonstrate the enterprise’s intellectual input.
- Never ignore the post-review. Any AI-generated content shall undergo manual review before being released, with special attention to the authenticity and ethics of the advertisement, as well as the presence of explicit or implicit labels.
- To strengthen awareness of copyright and data security. Employees or suppliers using AI software should be reminded to ensure that the content “fed” to the AI tool does not infringe the rights of others, and the uploaded information should not contain trade secrets, personal information, etc., so as to comply with the requirements in the field of data security.