The rapid development of new technologies, especially artificial intelligence, has provided users with the space to create new products and to automate processes. At the same time, the widespread use of tools has led to a loophole in applicable law.
Numerous questions arise, including who should be held liable for a wrongdoing (tort) committed by or with the aid of AI and whether AI should be granted a legal entity status to perform legal transactions, etc. Currently, a risk-based approach has been adopted to regulate what is still unknown. Without full knowledge of how and at what level AI will work, emerging legal regulations (including the AI Act, the EU regulation on artificial intelligence adopted by the European Parliament on 13 March 2024) focus on the negative effects resulting from the functioning of selected mechanisms by aligning the risk with obligations and restrictions.
In the case of infringements caused by AI, the most common one are those related to personal rights and copyrights.
Personal rights are commonly associated with the rights guaranteed to a natural person, such as human dignity or protection of the right to one's image. The Civil Code proposes an open catalogue of personal rights, however, it fails to provide a legal definition of this concept.
Although personal rights are most often linked to natural persons since they are individual and "personal" in character, this does not mean that they cannot also apply to legal persons, such as a business. An example is the company's image, which is understood as a set of features with which the company is associated by the public, and which affect an opinion about it and its reputation on the market. In judicial decisions, it is assumed that personal rights of legal persons may also include, among others, a name (business name), reputation, distinctive markings, or confidentiality of correspondence.
Violation of personal rights of the enterprise may entail serious legal and financial consequences and may affect the reputation established over the years. The dissemination of false information, the damaging of reputation, unfounded comparison with competitors or using a company's brand unlawfully may have real consequences for the entity concerned. A decline in trust among customers and contractors, a break or suspension of cooperation with business partners, problems to obtain funds, or a potential deterioration of the competitive position on the market are only some of image-related and business consequences that an enterprise may suffer when its personal rights have been violated.
As artificial intelligence develops, the concept of personal rights as belonging to both natural and legal persons is expanding. To be able to generate content corresponding to user-generated prompts, AI systems must be fed with as much data as possible. These materials often contain works protected by copyright law as well as those that may violate personal rights of specific entities. Therefore, it may happen that an AI model will generate content presenting a company's image in a negative light or using correspondence that has been leaked through the improper use of electronic tools. The mere use of the company's "image" is not prohibited as it is present in the public domain, similarly to the image of a public figure as set out in the Act on Copyright and Related Rights. On the other hand, portraying the enterprise negatively may constitute a violation of personal rights, particularly given that AI chatbots tend to "hallucinate". In addition, this increases the risk of misrepresenting and providing misleading information about the activities or persons associated with the enterprise.
At this point, the question arises as to who is liable for such a breach. So far, actions have been brought, depending on special regulations, against a natural or legal person, or an organizational entity without legal personality.1 If the AI-generated content infringes third-party rights, can AI be granted legal personality and can it be held liable for wrongdoings? Opinions vary, but currently the European Union legislation attempts to prevent artificial intelligence from being legally equated with humans. By definition, AI is intended to complement human work and activities, and not to be a human substitute, which is a threat rather than help.2
The same applies in the context of copyright law. In various countries, legal decisions in this area are evolving. However, the prevailing opinion is that artificial intelligence cannot be the creator since a work should reflect the author's personality and their creative choices. Although AI generates content based on what it "saw," it makes certain choices when creating a new product. These choices are dictated by algorithms and mechanical selection of the best elements. However, work is underway to develop machines with cognitive abilities, enabling them to perform activities they have never been taught before. There are far-reaching theses as to whether in such a case the responsibility should not be attributed to the AI mechanism, which, like a human being, would make choices when creating a product using the information of its own "intellect."
Currently, AI cannot be considered an infringer as it has no legal personality. Under Polish law, in order to attribute civil liability for a tort, the provisions of the Civil Code should be applied, which in the case of AI is inadequate, because no regulation contained therein refers directly to AI and the Civil Code provides only an interpretative framework within which a specific provision can be applied. In this regard, considerations revolve around comparing responsibility for AI with "liability for a pet," and recognizing AI software either as a dangerous product, for which the manufacturer is primarily responsible, or as a "force of nature". All of the above theories are imperfect and need to be updated for realities in which artificial intelligence is used. When considering liability for the infringements caused by AI, an analogy with Roman law, which provided for liability for the damage caused by a slave, comes to mind. AI could potentially be treated as a "tool" of the owner or user, so whoever uses or is responsible for AI could be accounted for the damage caused by it. However, this also gives rise to some questions. Who should be held responsible: the AI creator or supplier who "taught" the AI model using protected data without prior verification or rather the user who commercialises an AI-generated product.
Until legal mechanisms are developed, it is advisable to protect your enterprise against potential violations by introducing an internal policy and regulations for using AI tools, auditing systems and training, as well as relevant agreements that specify the division of responsibility for the effects of operating and teaching the AI system. Under Polish law, the amendment to the Act on Copyright and Related Rights of September 2024 stipulates that the right holder can make a reservation for a given work to be excluded from TDM3 and that it is impossible to store works after such activities have been completed, i.e. after their purpose has been attained. Yet, it is not clear how this reservation should be made in practice to be effective.
Legal issues related to AI associated with, among others, infringements, particularly as regards personal rights and copyright, are playing an increasing important role in business transactions. As current legal regulations are inadequate in the face of technological development, it is up to entrepreneurs to take steps to eliminate legal risk most effectively.
Footnotes
1. The Code of Civil Procedure Act of 17 November 1964, consolidated text: Journal of Laws 2024, item 1568, as amended.
2.
3. Text Data Mining is an established type of automated analytical technique designed to identify certain interrelationships in large data sets that can then be reproduced in the material generated by AI systems.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.
