Context and challenges of generative AI in intellectual property matters

How should intellectual property, traditionally intended to protect works, be approached in relation to generative AI?

The rise of generative artificial intelligence is transforming content creation in many fields, whether text, images or video. These technologies, relying on machine learning models, make it possible to produce works on a large scale and in record time, thereby blurring the line between human creation and automation.

One of the major challenges of generative AI concerns the question of ownership of the rights in works produced by these technologies. Copyright, historically grounded in human contribution, raises the issue of attributing rights to creations generated by algorithms.

The evolution of generative AI not only complicates the protection of works under copyright, but also raises questions about patent and trademark protection. These intellectual property instruments, traditionally intended to protect inventions and commercial identities, must be reconsidered in light of new technologies.

Let's discuss your needs for 15 minutes!

Patents for AI-assisted inventions

The patentability of an invention generated by or with the help of an AI depends on the ability to demonstrate a human inventive contribution. In many cases, patent offices require that the invention result from an inventive process guided by a human, which limits the protection of inventions generated entirely by algorithms. Debates persist as to whether AI systems can be regarded as inventors and, consequently, whether patents should be granted to them.

Trademarks and protection of commercial identity

The use of generative AI tools in the creation of logos, slogans or other elements of visual identity may give rise to infringement risks if the resulting creations resemble existing trademarks. Companies must put in place control mechanisms to ensure that the generated content does not infringe third-party rights, while preserving the originality and uniqueness of their commercial identity. Managing trademark rights in a generative AI context requires heightened vigilance, particularly in the event of abusive or unauthorised use of the creations by competitors.

Common challenges for patents and trademarks

Recent case law shows a certain reluctance to extend protection to creations made without human intervention, underscoring the need to rethink existing legal frameworks.

Adapting the traditional criteria for protection (originality, inventiveness, distinctiveness) to works or inventions generated by AI remains a major challenge for legislators and courts.

I want reliable legal documents!

Legal liability associated with generative AI

The use of generative AI gives rise to liability questions in the event of infringement of intellectual property rights or production of unlawful content. The complexity of training, the use of protected data and the partial autonomy of the systems raise the question of liability, both from the standpoint of developers and that of end users.

Determining liability

Where protected content is used without authorisation to train or generate works, it must be determined whether liability lies with the provider of the tool, the user who configured the model, or a third party involved in the training process. The lack of legal personality of the AI places liability on the humans who intervene in its creation process.

Risks associated with databases and training

• AI models are trained on vast datasets that sometimes include protected works. If these data are used without authorisation or outside the scope of legal exceptions (such as text and data mining), liability may be incurred for copyright infringement.
• Contracts with AI providers should include warranties regarding the provenance of training data in order to limit the risk of litigation.

Exceptions and legal obligations

• The text and data mining exception allows, under certain conditions, the use of protected content for training purposes. However, the right holders' right to opt out must be respected, which further complicates the allocation of liability.
• The transparency obligations imposed by new regulations, such as the AI Act, aim to clarify the use of training data and to allocate liability in the event of a breach.

Generative artificial intelligence challenges the traditional foundations of intellectual property law. As technologies continue to evolve, legal frameworks must adapt in order to determine ownership of the rights in works produced by or with the assistance of AI. The recognition of human intervention as an essential criterion for copyright protection remains central, while prompting a rethinking of the protection of inventions and trademarks in an unprecedented technological context.

European and international initiatives, such as the AI Act and the adaptations of the DSM Directive, offer avenues for harmonising regulation. However, the debate remains open as to the balance between protecting rights and promoting innovation. For companies, this means putting in place clear contractual strategies and closely monitoring developments in case law and regulation in order to secure their creations and their business model.

This shifting legal landscape highlights the importance of close collaboration between creators, lawyers and regulators in defining standards suited to the challenges of generative AI.