Margarita Guerrero
Artificial Intelligence (AI), the technology that simulates the process of human intelligence through computational processes and machines, has experienced an unprecedented boom in the past years, impacting areas such as technology, healthcare, and finance. Intellectual Property (IP) frameworks have not been immune to such revolutionary changes as they also need to adapt to provide more clarity to creators regarding the use of AI in inventions and the invention or training of AI models.
For the European Union (EU), this topic has become focal as they have internalized the relevance of AI for ethics, innovation, and inventors’ rights. Hence, in 2024, the EU released an European AI Act that aims to regulate AI technology according to European values and standards. This Act sets a pivotal precedent for aligning AI technological advancements and outputs with IP policies.
In 2021, the EU introduced the first Act related to assessing and classifying AI developments in accordance with a risk matrix that provided guidelines for the inventions’ risk level. If an invention was determined as a high risk one, then it would need to go through mandatory third-party verifications. Even though the Act has helped to effectively govern AI systems by providing a framework for monitoring the autonomous growth of AI, it left remaining questions about IP, such as who would own the outputs of AI-generated works or how transparency can be ensured in AI systems that use proprietary databases.
Hence, the EU has provided more clarity for IP stakeholders in this new act touching upon some relevant topics. For example, regarding the risk matrix provided in 2021, this new Act clarified that AI technologies and developments classified as high-risk systems must have extensive explainability and proper documentation. These extensive documentation requests will provide IP regulators with more information and decision points to solve IP ownership disputes related to AI.
Another major development involved aligning the EU’s position regarding copyrights. Faithful to its historic philosophy of fostering inventions, the EU had been attempting to balance IP incentives for creators with the appropriate recognition of AI’s contributions to any invention. For example, if a creator could not be granted copyright protection for a brand he invented with a brainstorming process aided by a generative AI program, then there would be no incentive for the creator to boost its creativity process with AI. Hence, the EU agreed that AI-generated works without significant human intervention would not qualify for copyright protection. This perspective is in alignment with previous articles issued by the World Intellectual Property Organization (WIPO) that extensively discuss challenges of AI-generated inventions.
Another significant discussion from the act is the patent eligibility criteria considering the use of AI for inventions. The EU’s consensus was that there should be more transparency requirements to ensure full disclosure of the use of AI in the invention process. Consequently, this guideline may change the way patents are filed in the future, expecting larger examination processes and more extensive documentation for a patent filing. Moreover, this may also mean that failure to disclose AI’s involvement may lead to frequent patent disputes and invalidations.
The AI Act also touches upon the use of datasets in AI training. Since each AI model mimics human thinking, the creation and adaptation of each model involves training the AI with large amounts of data for it to use more learning resources. For example, if someone wants to train their AI model on DNA technological advancements, they will need to use a large volume of DNA research to train the AI, resorting to publicly available and restricted databases in many cases. Hence, the EU emphasized that any AI training should be done with lawful data access practices, which might bring challenges for inventors that were relying on copyrighted databases for training their algorithms.
Having discussed all the relevant provisions that this AI Act outlines, there are clear benefits for IP stakeholders navigating AI cases. The main one entails increased transparency for dealing with the intersection between AI and IP. With consensus in areas such as the use of databases in AI training and patent eligibility criteria for AI-aided inventions, IP regulators in the EU can provide more clarity to ambiguous cases involving the use of AI in any form.
Secondly, the AI Act heavily promotes ethical innovation, where it ensures that inventors have room and get IP protection for any invention, even if AI was involved in the process, while also promoting legal access to data and the responsible use and monitoring of AI. The EU sets high standards for transparency and accountability with respect to the use of AI technologies. While this is one of the first formal documents that clarifies IP regulations in the context of AI, this EU AI Act will be a strong example for other countries and regions of encouraging the responsible use of AI in inventions processes, rather than jumping to restricting it altogether.
Lastly, with this Act, the EU positions itself as a global leader in systematizing IP protection and AI governance. Even though the WIPO has issued guidelines and suggestions for IP lawmakers dealing with complex AI cases, the EU becomes the first region in the world to consolidate and formalize the ongoing discussions about AI regulation in an official Act. While suggestions have provided quick clarity for IP regulators, the need for clear frameworks is becoming more evident as AI technologies and the methods for developing such inventions are evolving unprecedentedly.
Artists and content creators may benefit from stricter rules of using copyrighted material in AI training models, ensuring that their inventors’ rights are protected. However, they will also face challenges if they were heavily relying on AI to produce artworks or content that they want to get copyrights for and where significant human intervention cannot be proven. IP experts and professionals in the EU will have the key role of enforcing these new measures, ensuring that they balance the fostering of innovation with the protection of inventors’ rights. Lastly, AI developers will have tighter transparency obligations, ensuring that they comply with the lawful access to data for training their created algorithms.
The EU’s future-ready approach to adapting the IP frameworks for AI regulation signifies a key step in dealing with recent challenges posed by this evolving technology. By fostering transparency, ethical access to data, and invention accountability, these measures will safeguard IP rights of inventors, promote innovation, and secure the fair development of AI models. While the EU refines its framework, collaboration with international bodies will be essential to harmonize global standards that address the intersection of AI and IP. Since this technology is borderless, common consensus with other regions regarding these regulations will be key for their effective enforcement. The EU AI Act will become a blueprint for other regions, shaping the future of AI governance across the globe.
References:
1. Marwood Group. (2024). A New European Regulatory Landscape for AI. Link
2. Nature. (2024). The European Union introduced the Artificial Intelligence Act. Link
3. WIPO. (2024). Artificial Intelligence and Intellectual Property. Link
Margarita Guerrero
Mguerrero@uhthoff.com.mx
