Impact of AI on Global IP Systems

Introduction

What is Artificial Intelligence?

Creating computer systems capable of carrying out activities that need human intellect is known as artificial intelligence or AI. AI analyses vast volumes of data to spot trends and conclude from the material it has gathered. This may be accomplished by using methods like robotics, computer vision, natural language processing, and machine learning. AI encompasses a wide variety of skills, including language understanding, learning, reasoning, perception, and problem-solving. The ultimate objective of artificial intelligence is to create robots that can mimic human skills and perform various activities with increased accuracy and efficiency. Artificial intelligence (AI) has the power to transform many facets of our daily lives completely.

Where do we see Artificial Intelligence?

With the development in technology and modern advancements, Artificial intelligence has been slowly integrated into the daily lifestyle of people, such as:

1. Virtual Personal Assistants: Personal Assistants in our phones or smart gadgets like Siri, Alexa, and Bixby are one of the common examples of AI. These assistants use algorithms and machine learning methods to process the input given and provide suitable outputs.
2. 2. Self-Driving Vehicles: In recent times the automobile industry has revolutionized itself with modern self-driving cars, which through different sensors take in inputs and take suitable actions upon the processing, such as the appropriate speed of the vehicle, etc. One of the common examples is Tesla.
3. Healthcare Sector: Artificial Intelligence has also taken up its space in the healthcare sector. The diagnosis of diseases and potential threats to life are now being computed using different artificial intelligence software. Artificial Intelligence helps the patient as well as the doctors in an enhanced diagnosis and more suitable treatment options.

Working of Artificial Intelligence:

Artificial intelligence is based on a set of instructions and codes that enable it to perform different tasks by providing it with decision-making capabilities like a human. The decision-making process involves a lot of steps which can be seen primarily as follows:

1. Collection of Primary Data: To make an intelligent decision, the artificial intelligence application needs primary data. The software upon having such primary data processes it using codes and algorithms. Thus, the primary step becomes the collection and gathering of primary data.
2. Selection of the suitable Algorithm: The particular task or issue that the AI system is intended to resolve informs the selection of AI algorithms. Different algorithms work well for different kinds of applications, including pattern recognition, regression, grouping, and classification. Neural networks, decision trees, support vector machines, and k-nearest neighbors are examples of common AI techniques.
3. Model Training: In order to teach AI models patterns and relationships, they are given either labeled data (supervised learning) or unlabeled data (unsupervised learning) during the training phase. In order to reduce errors and enhance its performance on the specified task, the model iteratively modifies its parameters throughout training. Neural network optimization methods like gradient descent and backpropagation are used in this procedure.

Overall, in general, artificial intelligence (AI) systems use data, algorithms, and computing power to learn from mistakes, make choices, and carry out activities on their own. An AI system’s specific operations are determined by its architecture, algorithms, and the type of tasks it is intended to complete.[1]

1.      AI and the Global IP System

We need a worldwide intellectual property (IP) structure that encourages innovation and invention if we are to benefit from generative AI. The fast uptake of novel technologies such as generative AI necessitates an adaptation of the IP management systems.

When the present intellectual property system was developed, innovation was more sluggish and concentrated on human creativity. It mainly promoted the advancement of technology that facilitated the manufacture and exchange of tangible goods. In order to make the IP system more usable, effective, and accessible, society must make sure that new technologies are used while also being aware of any potential concerns.

One of the main areas of intellectual property law development is the link between artificial intelligence and intellectual property rights (IPRs). Growing AI-related business activity, early case law, and legislative and international policy activities are making it more and more relevant in practice. The Swiss Intellectual Property Institute and Zurich University’s Centre for Intellectual Property and Competition Law are working together on a research and policy initiative about the future of IP law in the context of artificial intelligence.

This document outlines the AI/IP Research Project and offers preliminary policy suggestions for the creation of AI-related IP legislation. The recommendations address issues like AI authorship and inventorship in copyright and patent law, the necessity for specific rights to safeguard creative AI output, guidelines for assigning AI-related intellectual property rights, IP protection carve-outs to support the creation, testing, and training of AI systems, the use of AI tools by IP offices, and appropriate software protection and data usage policies.[2]

2.      Impact of AI in different aspects

2.1.  PATENTS

The arena of patents has evolved with time, and in contemporary times, the scope of subject matter that is patentable has also evolved, which in turn has modified the requirements of patents. As contemporary technology has developed, the patent system has faced fresh difficulties. Specifically, artificial intelligence (AI) technologies have opened up new avenues for invention that only minimally entail human intervention.

A pertinent question that arises is whether after the significant changes in patents and their requirements have gone, the current regime is capable of encouraging and rewarding innovation. AI is similar to previous computer-assisted inventions in several aspects. But it’s now evident that AI is capable of producing inventions on its own, and there have been multiple documented instances of patent applications where the person applying for a patent has recognized AI as the inventor.

2.1.1.      Inventorship and Ownership:

The process of invention has changed significantly as a result of the AI technologies’ quick development and increased computing capacity. With AI’s expanding ability to filter data, identify patterns, and make predictions, these technologies are working in a variety of sectors driven by innovation. These days, AI systems are advanced enough to provide results with very little assistance from humans. If such products were created by a human inventor, they could be eligible for patent protection. “A human inventor serves as the central figure in the design of the patent system. The main rationale behind patent law is to reward and encourage the creative actions of creators. Patent protection extends to creations that stem from human ingenuity, not just discoveries or clear extensions or modifications of previously known concepts”.[3] “The concept of ‘invention’ entails the ‘act of intellectual creation original to the inventor’, i.e. the mental act occurring in the mind of the inventor.”[4]

More precisely, “a human inventor serves as the focal point for the examination of issues on inventorship. Courts typically consider the concepts that the inventor had in mind when deciding when and by whom an invention was produced, or the notion of the creation.”[5]

When someone helps solve a specific problem, for example, their work is typically viewed by the courts as inventive, and they will be recognized as a (co)inventor. Conversely, an individual’s contribution will not be deemed inventive if it consists solely of an “unnecessary detail” or if it has management, administrative, or financial components[6]. As a result, the key factor in determining inventorship is the type of real contribution made to the invention’s conceptualization, which needs to be clever or innovative. Naturally, AI technologies cannot be credited as the invention in this way. “Regardless that AI activity may be instrumental, if not decisive, to the patentability of an invention and the latter’s success in solving a technical problem, these technologies—which, at least currently, are incapable of cognition—cannot be regarded as a deviser of the invention. Rather, it is important to view these technologies as a crucial tool in arriving at the invention.”[7]

2.1.2.      Condition of non-obviousness and AI:

For an invention to gain the protection of patent it must classify as an inventive step and be non-obvious. If an individual knowledgeable in the relevant art would find the invention obvious, that individual would meet the criterion for determining non-obviousness.

Although different legal systems have created distinct, organized methods for analyzing obviousness, the pertinent question is which comes in play is whether such invention will be “apparent or obvious” to a person who is already skilled in that art.

Artificial intelligence greatly broadens the scope of things that any normal human using such AI can find without excessive experimentation on his own, therefore the question of obviousness becomes imperative. A lot of inventions can be done or can be the outcome of extensive calculative ability which enables the artificial intelligence for “trial-and-error” searching. However, the outcomes may surprise a human with specialized knowledge if AI is not used.[8] Consequently, human capabilities are effectively measured against AI capabilities in the absence of appropriate modifications in the evaluation of this kind of technology. This establishes a very low bar for obviousness, which might make most inventions produced by AI invisible to an expert in the field who only needs to use their mental faculties and common knowledge.

Determining the invention’s pertinent field of application and the extent of the prior art is another important topic. “Hyposing what would have been obvious at the priority date to the person skilled in the art to which the patent in suit relates” is how obviousness is determined.[9] Determining the “field of endeavor in which the inventors were working” is crucial, and it’s best to avoid “both unduly wide and unduly restrictive definitions” because they may lead to problems.

The selection and evaluation of the previous art are aided further by the appropriately designated relevant art. Specifically, “in evaluating obviousness, the courts consider the state of the art from the perspective of the person experienced in the art, who is only assumed to have examined the data accessible in their own or closely related fields—thus limiting the potentially extremely large reach of the subject.”[10]

2.2.  COPYRIGHTS

AI programs can create creative and literary works on their own. This ability of AI has raised a lot of concerns pertinent to copyrights, as copyright is linked to the creativity of humanity and aims to protect, promote, and ultimately acknowledge such creativity. If the copyright system fails to cover the creative work produced by the AI application, then it will be viewed as a system that aims to promote the creativity of humans more than the machines. On the contrary, if the creative work by Artificial Intelligence becomes eligible for the copyright system, then it will mean that the value of creativity either by the machine or by a human has equal value, and the source of such creative work will be meaningless but only the work will be valued.

In contemporary times, businesses in different fields such as the gaming industry, music industry, journalism, etc., use Artificial Intelligence for the creation of content. However, if these creations are not made eligible for the protection of copyright, then it would mean that these creations are freely available to use and anyone can use them without any restrictions. This free availability of the created content will pose a serious threat to the different industries and businesses using artificial intelligence in running their business, thus law needs to be made in light of this situation.

A report by the European Commission on IP and AI[11] has stated that “we could be moving towards AI autonomy, at least to a level that the human contribution is trivial to the creative or inventive process and therefore we could be entering into an era where machines will not only assist humans in the creative process but create or invent all by themselves. However, we are not presently at that stage, and at present AI technology is not currently truly autonomous”. When we talk about the technology being truly autonomous, it would mean that throughout the process of creation by artificial intelligence, almost no human intervention is needed but this is not true.

2.3.  DESIGNS

Artificial Intelligence is also capable of creating designs on its own, akin to the inventions made.

When we talk about the creation of designs by Artificial Intelligence software, Computer Aided Designs come into the picture and have been present for years now and they don’t seem to pose a challenge to the policies. Design rights are meant to safeguard a product’s appearance. Copyright and other intellectual property rights may intersect with design rights. According to WIPO, “the ornamental aspect of an article constitutes an industrial design in a legal sense. An industrial design may include two-dimensional elements like patterns, lines, or color, or three-dimensional elements like an object’s shape.”[12]

Certain AI programs can develop designs on their own, which are also known as “AI-generated designs.” “On the other hand, other AI applications are limited to assisting in the generation of designs, which are referred to as AI-assisted designs, which are a type of computer-aided design”.[13] For the determination of eligibility of an AI design for the grant of design protection it becomes necessary to evaluate if a skilled, deliberate act was necessary to develop the design. For instance, a design produced by a meticulously built AI system that is fed the right data for a certain goal may qualify for design protection. Users of the Artificial Intelligence software may also provide their inputs relating to the design and visual appearance of the AI application, which will allow the application to produce a design by itself by taking into consideration the inputs provided by the user. To promote legal clarity, it seems most reasonable to implement particular legal requirements that govern the ownership of independently (autonomously) developed AI designs. However, the grant of design protection will depend on the case.

2.4. AI AND TRADEMARK

Trademarks give brand owners exclusive rights by identifying the source of goods or services. The likelihood of trademark infringement rises with the increasing prevalence of AI-generated material. Artificial intelligence algorithms may unintentionally produce content that violates registered trademarks, creating confusion in the marketplace.

Artificial Intelligence software and applications are capable enough to produce slogans, logos, and brand names that mimic recognized trademarks. An AI might be used to propagate false information or spam using those created symbols, seriously hurting the owner of the trademark.

The confusion and infringement brought about by AI-generated content could have major repercussions for trademark owners. Using a trademark to protect one’s goods and brand serves the objective of stopping others from making similar things. Now, all it takes to replicate the diligent effort of a trademark owner is a few AI commands.

TOOLS DEVELOPED BY WIPO

• International Patent Classification-CAT

WIPO created the International Patent Classification-CAT (IPCCAT) initiative, which replicates previously registered patent classification processes to apply AI to the categorization of patent documents in a particular context. Using neural network technologies, applicants and patent examiners from various initial public offerings (IPOs) can more easily categorize pertinent patents into technical units according to their class, subclass, and groupings under the International Patent Classification.

There are other situations in which this technology can be used, such as document evaluation, analysis, and extraction based on predetermined patterns, therefore this is not their sole application. Additionally, this use could be extended to a wide range of other domains, such as trademarks, which have been studied in the specific context of AI for the Nice Classification (“NCL”).

• AI-powered image search

The introduction of a deep neural network and cutting-edge technology using figurative elements categorization data is another intriguing study being carried out by the WIPO. Its purpose is to detect similarities between colors and shapes of unique signs.
Recall that specific colors and shapes (like Tiffany Blue) as well as figurative components like the Instagram logo are also protected by trademarks.

• Translation tool:

A free translation tool called WIPO translate can be found on PATENTSCOPE. With the help of this incredibly precise technology, patent paperwork may be translated into ten different languages using sophisticated neural networks.

It makes translations easy for end users, IP experts, IPOs, and other national bodies that require them. Thus, it can be said that these kinds of technologies save costs while promoting innovation, knowledge dissemination, and the removal of linguistic barriers.

CONFIDENTIAL INFORMATION AND TRADE SECRETS

The legislation may offer more adaptable protection for trade secrets or private information. The data that is used to train AI systems, their operational methods, and the way they function might all be covered by this kind of protection. Legislative measures on trade secrets coexist with a common law system of confidentiality protection in the United Kingdom. Contrary to copyright, trade secret protection in the EU is extensively harmonised and protects against illegal misappropriation, so long as appropriate safeguards are put in place to maintain the confidentiality of the knowledge.[14]

What responsibilities have arisen?

• Developers

Long-term data sourcing will need initiative on the part of AI developers and investors to know the primary source of data being used. The “LAION-5B dataset”, is known to contain a significant amount of copyrighted works and has about 6 billion annotated photos gathered from indiscriminate web scraping and it serves as a foundation for models like Stable Diffusion, Midjourney, etc.

To improve transparency regarding the works included in the training data, developers should focus on ways to preserve the provenance of AI-generated material. This would entail documenting the content development platform, the settings that were utilized, the metadata of the seed data, and the particular prompt that was used in the creation of the content. Such information would not only make it possible to reproduce the image and easily verify its authenticity, but it would also provide evidence of the user’s intent, protecting business users from potential intellectual property infringement lawsuits and proving that the output was not the result of deliberate theft.

• Creators

Individual content producers and content-producing brands should take precautions to safeguard and assess the danger to their portfolios of intellectual property. This entails actively searching for their work, including visual components like logos and artwork and textual parts like image tags, in assembled datasets or massive data lakes. Although terabytes or petabytes of content data would be too much to handle manually, current search tools should enable the efficient automation of this process. Even concealment from these techniques can be promised by new tools.

Content producers should keep a close eye on social media and digital platforms for the emergence of works that could be derivative of theirs. At times the essential elements such as the logo of a brand or some distinct colour or pattern might not be present in the work produced by the AI application. Still, the presence of other salient elements may also indicate that the content of a particular business or creator was used in the creation of the result. This similarity in the result produced may indicate the intention of using the goodwill of that company or creator by using such identifiable elements.

Conclusion

The rapid advancement of artificial intelligence (AI) is reshaping the global intellectual property (IP) landscape, creating both opportunities and challenges. As AI-generated works blur the lines of authorship and ownership, existing IP frameworks face significant tests, calling for responsive legal adaptations.

AI’s ability to autonomously create content raises important questions regarding rights attribution. Some jurisdictions are considering reforms to recognize AI as a co-creator, while others retain a human-centric focus, highlighting the need for international consistency to avoid legal uncertainties. AI also enhances IP management through improved patent searches and automated copyright enforcement, increasing efficiency and reducing infringement risks. However, it also poses risks, such as the creation of counterfeits and deepfakes, which challenge enforcement efforts.

The future of the global IP system will rely on balancing innovation with rights protection. Policymakers, legal experts, and technology leaders must work together to develop frameworks that support AI advancements while addressing ethical concerns. Key elements will include transparency, accountability, and fair benefit-sharing.

To succeed in this evolving environment, the IP community must remain flexible, developing policies that protect human creativity and embrace AI’s potential, ensuring that the global IP system continues to foster innovation.

[1] “What is Artificial Intelligence?, 19 Mar, 2024, at https://www.geeksforgeeks.org/what-is-artificial-intelligence/#what-is-artificial-intelligence (accessed on 6^th June, 2024)”

[2] “Picht, P.G., Thouvenin, F. , AI and IP Theory to Policy and Back Again – Policy and Research Recommendations at the Intersection of Artificial Intelligence and Intellectual Property. At https://link.springer.com/article/10.1007/s40319-023-01344-5 (accessed on 6^th June, 2024)”

[3] “PLG Research Ltd v Ardon International Ltd [1995], FSR 116, 137

[4] Robert L Harmon, Harmon on Patents. Black-Letter Law and Commentary (BNA Books, 2007) 32

[5] Burroughs Wellcome v Barr Labs, Inc., 40 F.3d 1223, 1227 (Fed. Cir. 1994).

[6] IDA v University of Southampton [2006] EWCA Civ 145, [39]

[7] Noam Shemtov, A Study on Inventorship in Inventions Involving AI Activity (EPO, 2019) 19.”

[8] “Lisa Vertisky, ‘Thinking Machines and Patent Law’ in Barfield et al (eds.), Research Handbook on the Law of Artificial Intelligence (Edward Elgar, 2018) 497.

[9] Technograph Printed Circuit Ltd v Mills & Rocky [1972] R.P.C. 346 at p. 362

[10] Lionel Bently et al., Intellectual Property (5th edn, OUP 2018) 584”

[11] “Maria Iglesias, Sharon Shamuilia and Amanda Anderberg, Artificial Intelligence and Intellectual Property – A Literature Review, EUR 30017 EN (Publications Office of the European Union, 2019).”

[12] “WIPO, ‘Industrial Designs’, at https://www.wipo.int/designs/en/, (accessed on 7^th June, 2024)

[13] WIPO Secretariat, ‘Draft Issues Papers on Intellectual Property Policy and Artificial Intelligence’ WIPO/IP/AI/2/GE/20/1, at https://www.wipo.int/edocs/mdocs/mdocs/en/wipo_ip_ai_ge_20/wipo_ip_ai_2_ge_20_1.pdf , (accessed on 7^th June, 2024)”

[14] “Aaron Hayward, Anna Vandervliet, 23 Mar 2023, The IP in AI: Can IP rights protect AI systems?, at https://www.herbertsmithfreehills.com/insights/2023-03/the-ip-in-ai-can-ip-rights-protect-ai-systems , (accessed on 10^th June, 2024)”