Loading...
topic
HOME  /  PUBLICATIONS  /  UNFOLDING THE DYNAMICS OF IP LAW IN THE EMERGING AI – DRIVEN ERA

Publications

Search Publications
Unfolding the dynamics of IP law in the emerging AI – driven era

Unfolding the dynamics of IP law in the emerging AI – driven era

The outstanding achievements of generative AI offer remarkable benefits, but at the same time pose tremendous legal challenges that organisations need to be aware of. As AI tools and projects, such as ChatGPT, Dall and others, become increasingly proficient at generating content with minimal human involvement, the concerns on intellectual property rights are growing exponentially. This insight aims to delve into the future implications of AI-generated works, touching on ownership, protection, and the potential changes required in intellectual property laws to accommodate these technological advancements.

Traditionally, intellectual property laws have centered around human creativity, but with AI's growing role as a content creator, the landscape is evolving. The question of what work is and what is not protected under law is often a complicated one, with the answer almost always dependent on regional rules – this is to say that, securing copyright protection for AI generated works relies on existing IP laws. Nonetheless, existing legal frameworks vary among different legal systems. In the EU for example, ownership is exclusively delegated to humans, contrary to the UK, which accepts and provides intellectual property rights to computer-generated works.

In cases where AI is more than just a tool in the creation process, questions on the ownership of the AI-created work arise almost instantly. In situations where a human author is absent and the work constitutes an AI-generated creation, authorship is delegated to the person who took all the necessary steps for its creation. According to UK copyright law, artistic works created by a computer can benefit from copyright protection in the UK. In contrast, the EU seems to take a different approach as it does not consider computer generated works justifiable for copyright protection. In fact, the ECJ in the cases of Infopaq, Bezpečnostní softwarová asociace, Painer and Football Dataco adopted an author’s-rights-based-approach in that it interpreted copyright originality as deriving from and being attached to an author’s own intellectual creation. Organisations should think out of the box by indulging in innovative ways to secure copyright protection for AI generated works - this may include coming across applicable law and implementing strategies such as securing contracts to further their business objectives whilst at the same time, securing desirable IP protection (e.g., they must assess whether AI inventions are eligible for patent protection and seek the countries they want to grant one).

The UKIPO treats AI inventions similarly, as they must satisfy the same patent eligibility criteria. Any patent application must identify the inventor who currently must be a natural person. Accordingly, an AI system cannot be considered as an inventor, nor the owner of an AI machine could be entitled to a patent for a machine-made invention by stating that the machine was the inventor. Relevant patent applications from AI machines rejected by European Patent Office (EPO), emphasizing that only humans can be inventors on a patent application. Hence, where patentability requirements cannot be met, the element of trade secrets may be more efficient for organisations to secure data protection.

Copyrights are delegated to original works, requiring work to be an author’s own intellectual creation to meet the criterion of originality. When the discussion comes to infringement of copyright, it must be clear that there is a work protected by copyright that might me infringed. The potential for copyright infringement in the realm of AI begins with the training of data inputs. Organisations must familiarize themselves with applicable laws and agreements governing generative AI usage. It is important to carefully assess the source of input data to ensure compliance with proper licensing and consent. The legality of data inputs is crucial to avoid copyright infringement, and organisations should take proactive measures to guarantee that they are not using third party works without necessary permissions.

Moving beyond data inputs, the outputs generated by AI systems also raise concerns on potential copyright infringement. As AI tools draw from and produce content based on vast datasets, there is a risk that the outputs may inadvertently infringe existing copyrights. The question then arises of who is going to be held liable for copyright infringement by AI generated works? By way of an indication, based on the terms and conditions of Chat GPT, by delegating any rights from the inputs and outputs of AI to the user, some argue that liability for copyright infringement should exclusively be conferred to the user of the AI system. On the other hand, given the possibilities of AI, others suggest the adaptation of AI tools for the investigation and identification of any works that may infringe copyright. Whatever the case may be, the current liability and infringement landscape demands stakeholders to be vigilant (on an independent and active basis) in understanding and adhering to intellectual property laws.

Stakeholders must take proactive steps if they wish to ensure effective IP protection in the AI landscape. Users of AI tools should inquire about the models being trained on protected content, review terms of service and privacy policies, and avoid generative AI tools that do not confirm proper licensing of training data by content creators. Likewise, AI developers must ensure compliance by acquiring data through legal means, including licensing and compensating IP owners for the use of their works in training data. Also, businesses should evaluate transaction terms, demand terms of service from AI platforms to confirm proper licenses for training data and include disclosures in contracts to protect intellectual property rights.

The creation of generative AI checklists for contract modifications can help reduce unintended risks associated with AI use and in the matter of content creators, they may consider building their own datasets to train AI platforms with lawfully sourced content, ensuring that their intellectual property is protected whilst minimizing the risk of copyright infringement.

In times of rapid technological advancement, legal frameworks are undergoing profound changes spurred by innovation and creativity. The future may see a shift in laws, acknowledging AI entities as authors of creative works. As, AI’s role in content generation evolves, there is a growing anticipation for legal adjustments that align with the unique contributions of artificial intelligence to the creative landscape.

The proposed EU AI law stands at the forefront of these changes, representing a comprehensive legal framework within the EU. This landmark legislation is expected to set a precedent, providing guidance on the intricate intersection of AI and intellectual property. As the EU leads the way in establishing consistent standards for AI, these recent legal innovations may serve as a blueprint, influencing global conversations on the challenges and solutions related to AI and IP protection.

Related Publications

Foundations are institutions intended to serve non-profit purposes and are established for the promotion of specific objectives outlined in law, including the promotion of commerce, art, science, religion, ...
On the 30th of November 2023, the naturalisation process administered by the Civil Registry Law in the Republic of Cyprus underwent a key amendment, specifically for high skilled foreign nationals working ...
A provisional agreement was reached upon after a final negotiation round between the EU Parliament, EU Council, and EU Commission meaning this milestone legislation is nearing implementation, making it ...
Artificial Intelligence otherwise commonly referred to as “AI” is a field of computer science that mainly focuses on creating models that use cognitive intelligence to perform tasks which were (and, to ...
Our firm has authored the Cyprus chapter of Lexology Getting The Deal Through: Asset Recovery 2024.