Generative AI has an intellectual property problem, here's why

Felicity Hammond Associate felicity.hammond@bsabh.com

Munir Suboh and Felicity Hammond of the BSA Intellectual Property team recently spoke with Arabian Business regarding artificial intelligence (AI), answering the following questions:  

How would you describe the current state of intellectual property rights in relation to generative AI and its impact on AI-generated content? Does generative AI have an intellectual property problem?

Generative AI is a type of advanced capabilities AI in comparison to the Narrow AI. It can create original content based on the patterns it has learnt from accumulated data, information and other features. Whilst the Narrow AI is known as “Weak AI” as it can help only in limited tasks, Generative AI is a result of deep learning techniques which allow it to decipher more complex patterns and make its own decisions and perform more advanced, or unlimited tasks, without human intervention or direct role. However, Generative AI is still manageable and has not yet reached to suppress human intelligence, create emotions, beliefs, needs or desires of its own, which is referred to as “Super AI capabilities”, which remains hypothetical.

The main challenge associated with Generative AI is determining who owns the produced intellectual property from such capabilities. Due to the ‘deep learning’ involved in Generative AI and different directions in compiling data and outcomes, there is not a direct human intervention or link to existing data, information or rights created by third parties, especially where the content is produced as a result of the systems own deep learning. This indeed blurs the lines as to who owns the IP of the Generative content and the issue will create many challenges.

Since IP was created to protect human creativity, innovation and advancement of science and work of arts originality, the absence of direct human intervention for the creation of a work has not been explicitly anticipated in the existing legal frameworks and to date. In term of liability, the laws made a link between the outputs and implications of using machines and liability under tort theory, however, the ownership of rights remains an issue to determine and still to be addressed in the laws. To date, there is no universally established guidelines on how to treat AI-generated content. In the past, any tools of the creator were treated merely as tools and the reasoning behind this is that the creator was still responsible for the creative input associated with a work. Most recently however, we saw legislations in EU and proposed draft of IP law in some countries, such as KSA, to deal with the copyrights resulted form AI. The proposition to make the products in public domain if there is no clear link to preowned machine or an entity that controls/owns such machines.

As AI technology advances, further discussions and developments in intellectual property laws will take place to address the emerging issues and provide clearer guidelines for the protection and management of AI-generated content.

How do existing intellectual property laws address the issue of ownership and protection for AI-generated artwork that gains critical acclaim and commercial success?
It is possible to allege application of the existing intellectual property laws, namely copyrights and/or patents, on any innovation or original works. However, the applicability of such protection is largely dependent on context and outputs and whether the same can protected or not. For example, in a situation where a human author contributed to the work with the support of AI technology (or with AI as their tool) it could be said that they are the creator of the work as they are responsible for the creative input as well as the machines/tools that are being used to come up with the new work. AI will have the analogy of supportive tools that are being used by traditional innovators/authors, such as cameras for photographers, painting brush by painters, computer by software coders….etc.

Therefore, and in a situation where a human trained an AI machine, they could be said to own any ‘work’ that it produces. The AI however should not create or give a support that is legal questionable and therefore the manufacturing of AI should be safely programmed. There may also be situations where AI arguably generated content is itself a derivative work and in these circumstances, the original owner of the copyright would need to be consulted and may receive royalties for this when the original work is being used in the derivative work. For instance, the AI that helps programs to generate or advance video games stages deprived from a story written by an author, the video game is going to be considered as derivative work created by the programmer(s) who use AI to produce the content.

How should the legal system navigate granting copyright protection to AI-generated content (art, music, text, etc) that imitates the style of renowned artists or draws inspiration from copyrighted works?
Successful navigation of this is dependent on the level of imitation and/or inspiration that the AI tool draws. When copying happens, the user of AI will be liable for such imitation and will have to cope with the infringement claims that may be pursued.

Law makers will need to introduce new clauses or enact a new law to cope with many new issues such as establishing or revising existing creativity criteria, originality, assessing the level of creativity involved in AI-generated contents when compared to pre-existing contents, define plagiarism in AI content…etc. Furthermore, Law makers, perhaps should deal with the moral rights issue that could be questioned when AI generates works to its users. Right of attribution, right of revision, right of alteration and/or distortion are among many other rights that law should specify how practically AI generated contents can be controlled or shall react when such a claim is made.

In cases where human programmers create the algorithm and AI systems autonomously produce artwork, how should intellectual property rights be assigned?
To date, there is no agreed interpretation of how this should be handled so parties should follow the traditional way to assign such rights. A human should be involved or linked to the machines that produced the artwork and should assign or transfer such rights when the output created by the AI system are identifiable and can be determined. For instance, the human programs created algorithm and AI system should be the person to assign such rights and works, such as codes, programs, designs in computer program, …..etc. The logic of this is that the law only allows assignment from natural or legal entity that has human who can represent the same, such as director/owner. However, as AI systems become more intelligent and increasingly autonomous, new legal frameworks are highly needed and required to account for the lack of creative human input. The Saudi IP law that is being drafted and shared with public for consultation dealt with this and considered that IP rights resulting from machines that has no human input or link to such machine will be fallen within the public domain, i.e. no protection and can be utilized and used by natural or legal entities. Similarly, the UAE was among the first states to setup a standalone Ministry to deal with AI, called Ministry of Artificial Intelligence, a few years ago to enhancing the government performance levels by investing in the latest technologies and tools of artificial intelligence and applying them in various sectors. We still need to see how the IP laws will adopt or deal with the outputs of AI and the effect of this on ownership, enforcement and protection.

How should the legal system address the production and commercialization of AI-generated content mimicking the style of deceased artists, considering the lack of consent?
If we park on the side the issues resulting from right of publicity and/or privacy, we will still need to deal with the question on whether a copyright infringement happens or not. In fact, Deceased artists inherit their copyrights to their heirs and the same remain protected for 50 years after the death. Thereafter, the copyrights fall within public domain. Moral rights however remain protected permanently and can be exercised for any person that proves capacity to pursue protection on such rights. Mimicking the style of deceased artists should be assessed based on whether such art works are still protected or not. If the same are in public domain name, copying or mimicking such works by AI or use or exploit or utilize such work should not create any liabilities as a general rule save the moral rights element and consideration. However, if the copied works are still within the protection period, the exploitation of such rights within a consent from the heirs or holders of rights may result to infringement claim or liability on AI’s user or machine holder.

How can copyright holders effectively protect their rights in a landscape where generative AI algorithms create and distribute an unlimited number of potentially infringing derivative works?
Copyright holders should identify the potential entities or individuals that responsible, controlling or managing such AIs and move to enforce their rights against derivative works that utilize or use its original works without a consent.

In practice, many experts indicated that watermarking and metadata are among many available tools that can be used to technically deter and prevent the unauthorised use of copyright works by third parties, i.e. controlling AI systems or algorithms that create derivative works from violating the rights of original copyright creator. Furthermore, the use of digital right management technology to prevent screen grabs, unauthorised access and use of content is also a useful tool for owners, especially in an online context.

What legal reforms or adjustments do you believe are necessary to ensure a fair and equitable intellectual property framework that appropriately addresses the challenges posed by generative AI?
Due to the global nature of AI technology, universal cooperation on AI laws and standards is required and this can be completed by signing treaties and conventions. Without harmonised standards, protection will be inconsistent and as a result, difficult to harmonize between the systems and ultimately create challenges in enforcement of IP rights. Berne convention was a good tool last century to bring harmonized copyright protection and a similar proposition/framework should be considered to control or regularize the outcome of AI and its implications on IP rights.

Furthermore, AI functions proved to be very useful for human and science developments and this can continue to support researches and developers. The General AI is showing one of the useful outputs from AI, and the Super AI, if its not going to remain as a hypothetical concept, will certainly be of a great help to progress and advance many objectives. We are now thinking of equitable and fair proposition to reward the humanity from creating and advancing the AI functions but we should expect in two or three decades that AI will require a fair legislative platform that reward their productivity and innovation. Something was raised or suggested in many occasions by experts who are very optimism to see Super AI capabilities in our time.

This Q&A formed part of a larger piece by Arabian Business, which can be read here: ChatGPT, Midjourney: Generative AI has an intellectual property problem, here's why.