AI, IPR, and Digital Sovereignty: Why Local Regulation Matters

Introduction:

The EU AI Act 2024, which has been in the works since 2021, represents a milestone in regulating AI. It is among the first acts that establish enforceable and detailed frameworks regarding AI. The act has taken into consideration any feedback which it had received from its drafts, and most of its provisions are set to come into force two years from now.

The scope of the EU AI Act 2024 is comprehensive with a horizontal effect, making it wide in coverage, from all sorts of AI activities down to entities dealing in AI. This act is poised to have a strong impact on companies operating in the European Union. It applies to EU countries but is not a replacement for the GDPR. Because of Brexit, it does not include the UK.[1]

The EU AI Act 2024 contains a separate chapter for the risk-based approach of general purpose AI, further classifying AI into unacceptable, high, and low-risk categories (Arts. 2(3), 2(4), 2(6), 2(8), 2(10) and 2(12)), thus showing the level of risk to be identified. Unacceptable risk involves a highly debated point of conflict in the European Parliament: to regulate or not to regulate biometric scans in public spaces. France felt that from the point of view of security, biometric scanning was important, while others felt it was a breach of privacy, in particular, in public places. Still, later, it was considered an unacceptable risk.

It is imperative to know what high-risk AI is and what comes under its ambit. So, general-purpose AI, such as ChatGPT, Gemini AI, and Co-Pilot, could be used for a variety of purposes and under a plethora of sectors, such as scanning CVs, healthcare, education, etc., and this could pave the way for racial or gender-based cognitive biases. Again, this is emphasized upon under Art. 6(3) EU AI Act, which states that areas will be considered high-risk if they include profiling of people. [2]

Still, here it shifts all onus on the deployer and none on the end user, absolving the user of almost all liability (Art. 6(3) EU AI Act). The approach employed with General Purpose AI is more user-based and categorizes it as high or low risk. Open to interpretation, section 4b of the act says it may be used as high-risk AI systems. Solidifying that generative AI's categorization depends on how it is used.

 Coming to limited-risk AI, which imposes transparency obligations on the deployers, these systems generally interact with humans: chatbots, biometric scans, and deep fakes. These do not necessarily have to conform to any additional legal obligations. To make sure this categorization is applicable, the EU follows a sandbox approach in which the member states or the Data Protection Supervisor oversee the AI systems before they come into the market.

The current risk-based approach provides a great structure to work upon. It ensures a high level of applicability and, through the sandboxing approach, [3]makes sure the upcoming AI models conform to the established framework. Noncompliance on the part of the deployers attracts a heavy penalty ranging from 1.5% to 7% of a firm's global sales turnover, depending upon the severity of the offense and the size of the company. The act also creates a complaint forum for citizens under Art. 84, thus providing an important avenue for redress and a guarantee that the rights of the citizens are protected.

The EU has made it necessary that "the Union must oversee fundamental rights, guarantee 'human oversight,' and keep data governance and cybersecurity for high-risk AI systems." One important implication of the AI act of the Union involves emphasis on transparency regarding AI regulations.[4]

AI and IPR:

The intersection of artificial intelligence and intellectual property rights is an example of a complex and emergent issue in international law. The TRIPS agreement that was developed by the WTO in 1994 set a global standard aimed at the  protection and implementation of intellectual property With the gradual evolution of the AI technology into generating vast volumes of creative output, such as in visual art, poetry, and music, we finally see that the current IPR paradigm of TRIPS fails to address these novel works of creation.

The Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS) lays down some minimum requirements that the protection of intellectual property should observe in its member states, which include patents, copyrights, trademarks, and geographic indications. However, its stipulations still lack clear definitions and regulatory directions on the ownership, protection, and enforcement of intellectual property rights that relate to works, which are autonomously created through artificial intelligence systems. In its turn, this gap creates a high level of uncertainty and inconsistency in how the works created with the help of AI are judged in accordance with the domestic legal regulations.[5]

As the technology of artificial intelligence evolves at a  rapid pace, these systems have been able to create creative content which would otherwise be part of the intellectual property domain. An example of this is that AI algorithms are capable of creating visual artworks with complex designs, writing poetry that explores poignant subjects in depth and creating musical compositions that, in fact, surpass those of famous human composers. The new deliverables, however, instigate some fundamental questions on the applicability of the prevalent intellectual property laws.

Traditionally, intellectual property rights have been based on human creativity and authorship. The right to copyright is usually granted to the persons or organizations which are the source of original works of authorship. Producing artwork or a composition of music or music autonomously by an artificial intelligence system effectively disputes the premise of human authorship on which the traditional concept of copyright law is based.

Since there exists no comprehensive regulatory framework of AI-generated works in the context of the TRIPS Agreement, the heterogeneous policy reactions of individual Member States has been taken. A minority of nations have adopted a liberal approach and made the intellectual-property protection of AI-generated work apply to products analogous to those created by human authors. Other jurisdictions have, on the other hand, taken a more conservative position, arguing that the intellectual-property right should protect creations and works that are based on human inventiveness and authorship. These conflicting stances create a discontinuous legal environment, with inconsistent results, thus creating substantive challenges to inter-border IP sales and enforcement.[6]

The issue of AI with regard to intellectual property goes beyond copyright into patentability. The traditional concept of patent is one in which the inventors are required to be human, while, with increasing advancement in the development of AI systems, these systems are more and more often contributing to new inventions. This raises the question of whether the legal definition of an inventor should be extended to include AI systems or whether patent rights are to be ascribed to the human operators or developers of the AI.

The Way Ahead: Making AI Locally Responsible:

The US is made up of 50 states, and the data privacy is different from that of the EU. It takes up a somewhat locally responsible role, and almost 18 states have their laws. While the US has not yet developed comprehensive legislation around AI, it helps to reduce friction and tailor regulations at a state level. Similarly, other countries should also seek to have laws that best fit their needs and requirements—locally responsible/globally responsible.

The US has what we call a 'fragmented approach', and the EU would have unified frameworks like the General Data Protection Regulation, or GDPR. The issues that may arise with the act are related to digital sovereignty: if the laws that the EU would have, every country would have of its own, this would strengthen data residency immensely-a vital part of digital sovereignty.[7]

At the same time, the barter existing where we give data to AI in exchange for better services is a kind of data protection. We need laws that are fit for countries to safeguard their data, which is hugely powerful and thus needs careful regulation. When the laws are locally responsible, they will be more effective because they will not be like an international treaty where there is almost no liability on the parties to the treaty in case of non-compliance.

The laws that would be drafted should aim to be specific compliance laws for AI, and when AI companies are held responsible for the use of their data, they will be more aware and really mindful of the use of the data being provided. If a company is told by a large part of the world that if it wants to be operative in our region, it should follow these rules, then the company will have to comply. The main idea behind making the AI act locally responsible is to enhance its effectiveness in order to keep it relevant.

Conclusion:

While artificial intelligence continuously evolves across the creative, economic, and administrative spectrums, regulation balances innovation, authorship, and accountability. So far, the EU AI Act has provided the basic elements of a risk-based framework, but the intellectual property systems, like TRIPS, remain unprepared to handle AI-generated outputs. Meanwhile, the growing relevance of digital sovereignty indicates that localized, contextually framed regulation of AI may prove to be more appropriate instead of universal models. Guaranteeing responsibility, transparency, and the protection of human creativity will remain focal in shaping the legal future of AI. Tags/Keywords: Artificial Intelligence Law, EU AI Act, TRIPS, Intellectual Property Rights, Digital Sovereignty, Patent Law, Copyright and AI, AI Policy