AI’s disruption in IP and evolution of jurisprudence so far

AI and IP: The Disruption and Evolving Jurisprudence

IP laws have, until now, in the absence of AI, contemplated ‘intellectual property’ to be the creation of a human-being’s intellect. This foundation has been disturbed by the recent explosion in AI, which has seen the creation of IP, from the so-called ‘intelligence’ of machines/software.

Like the present scenario for AI, IP regulation was sparse when IP was gaining recognition. While localized and sporadic regulation was being developed in different parts, eventually, most countries came together to agree upon a minimum standard of protection for IP, through the establishment of WIPO and execution of the TRIPS agreement. IP laws were introduced across the world, to incentivize creators by arming them with ownership, exclusive economic rights, and a period of protection, to enjoy a monopoly over monetizing their ‘property’.

AI, today, has advanced to a level where it is capable of self-learning and creating original content. AI-tools can generate art, write code, create unique royalty free music, write stories, compose poetry, etc.

For instance, ChatGPT can, on request, compose a scientifically accurate and original poem about collagen, in Shakespearean writing, within seconds. To achieve this, it ‘studies’ information fed into it about collagen and Shakespeare’s writing style, then composes poetry. Interestingly, when asked if the poem was copied or original, ChatGPT confirmed originality, and when questioned about content overlap, clarified that while it may contain certain elements similar to existing works, its content is original and unique to each user’s request. On ownership, ChatGPT states “As an AI-tool, I do not claim ownership of any content I generate. The text I produce is not protected by copyright laws and can be used freely without attribution.”

In contradiction, per OpenAI’s terms-of-use, all rights in ChatGPT-generated-content are assigned to the user, clarifying that ownership will, on creation, vest with OpenAI and then be transferred to the user. Certain other AI-tools clarify that companies who own them, will own content generated by them. The base issue here, for AI-tools, lies in the non-recognition of IP in AI-‘works’, like ChatGPT’s poem, as it wasn’t authored by a natural person. Most IP laws do not presently recognize AI as a ‘person’, which makes AI incapable of claiming authorship/inventorship. Besides causing disruption in IP norms, lack of protection for AI-works is likely to lead to misuse and ethical issues, if humans wrongfully protect and monetize AI-works as their IP.

There are presently no clear answers to glaring questions about authorship, ownership, protection and exploitation of AI-works. As law-makers attempt to provision for AI, adjudicating bodies, globally, are simultaneously being tasked with addressing novel issues around AI-works, and consequently, jurisprudence is developing slowly.

Japan, Canada, EU and India have recognized the need for legislation surrounding AI and IP. Most courts/IP offices have expressed that, if human creativity has played a part in the creation of work, using the assistance of AI-tools will not disentitle the human-author from protecting the work. US and Indian copyright offices have refused registration of works where AI-tools were named as author/co-author, whereas, Canada’s IP office granted copyright registration to an artwork where an AI-tool was named a co-author, and South Africa’s IP Commission granted a patent to DABUS, an AI-system, as an ‘inventor’.

DABUS, invented by Stephen Thaler, is the first-ever AI-system to independently create two inventions, for which Thaler filed patent applications naming DABUS as the ‘inventor’. These were rejected by the European, UK, NZ, Australian and US patent offices, since, per patent law, only a natural person can be an inventor. The implications of IP ownership by AI-tools and naming AI-tools as IP creators requires analysis, i.e., do AI-tools require IP ownership, when (as far as we know) they aren’t even capable of exercising rights as IP owners?

It is true that AI itself is a result of human creativity. AI-generated/created-works, however, have sparked a debate with many opposing views around ownership/whether they constitute IP. One view is, as there is always some level of human intervention at the backend of AI, AI-works are a result of human intellect. Another view is, AI-works can only be products of a combination of existing information, fed into AI-systems by humans, and cannot be original as they are incapable of meeting the requirement of ‘modicum of creativity’. A view in favour of AI-works is that advanced AI-tools use existing information only for analysis/machine-learning, and create entirely new/original AI-creations, with miniscule/no human involvement, and warrant protection.

Whichever view one subscribes to, an unregulated space is always replete with unanswered questions. For instance, who should own AI-works - the person who created the AI-tool, or the AI-tool’s user who provides unique prompts which generate AI-works, or the AI-tool itself? Wouldn’t automatic human ownership of AI-works contradict the fundamentals of IP laws? Is it ethical for humans to own/reap benefits of AI-works, with minimal creative involvement? Who owns AI-works created without human involvement, does their ownership remain undecided?

We find ourselves in a similar situation as the one faced by the world when IP required regulation. AI is here to stay and will continue to develop at a much faster pace than the law, unless the world’s countries come together, as they did before, and formulate a global framework to regulate and protect AI-creations.