The surge in the usage of GenAI tools, or more specifically, the capabilities that appear to be unleashed by the widespread usage of GenAI tools, has driven a range of concerns across the legal and policy spectrum, with the most litigation, worldwide, being driven in the realm of intellectual property (IP) law. And given that answers to these questions, especially on ownership of outputs, and on legality of training methods, are likely to have heavy impacts on who controls the infrastructure or direction of AI development, they have been heavily fought by stakeholders across spheres. However, unpacking this intersection of AI and IP in a conceptually useful manner requires us to look away from questions of private ownership and money, and revisit the idea of IP to start with.
Conceptual stretching and IP
To start with, “IP” does not actually refer to a single idea at all. Rather, it is an umbrella term that refers to a number of conceptually distinct ideas. The way Trade Secret policy may interact with AI development has vastly different consequences than the way, let’s say, Copyright policy would, or that Trademark policy would. For the purposes of the current conversation, it would perhaps be useful to refer more specifically to “exclusionary rights” – the legal entitlement to prevent others from replicating your creation – such as copyrights and patents. Why are exclusionary rights permitted, even encouraged, to exist in a free-market society? The reason is very specific: to acknowledge that there are situations where the gap between the ‘cost of creation’ ($$$$), and ‘cost of replication’ ($) is so wide that it may deter creators from making their creations public.
Why are exclusionary rights permitted, even encouraged, to exist in a free-market society? The reason is very specific: to acknowledge that there are situations where the gap between the ‘cost of creation’ ($$$$), and ‘cost of replication’ ($) is so wide that it may deter creators from making their creations public.
The first point to note here is that the primary goal is not to ‘reward’ but to ensure societies see more creations make their way to the public. The method is by rewarding with a temporary exclusion right.
The next point to note lies in the answer to the question of why it would cause this deterrence. Because in a market society, the ability to commoditize one’s works (‘protected’ or not) is often directly linked to one’s ability to receive recompense. And if the cost of creation ($$$$) is taking a significant amount of one’s time or resources, the assumption is that they need some way of recompense, or that creative/innovative activity would not take place. In other words, exclusionary rights exist to give ‘creators’ the potential of market power for their creations, via the right to exclude others from bringing replications to the market. (Akshat Agrawal expands on this very well in a new paper here). This is why, for instance, exceptions to exclusionary rights often permit personal usage or non-commercial usage or research-only usage. It is not about ‘control’ over the invention but about the limited ability to commoditize it, for a temporary period, so long as it fulfils certain criteria.
Once this is clear, it also becomes clear that the rationale for exclusion rights thus actually reduces as the gap between the Cost of Creation and Cost of Replication reduces.
Once this is clear, it also becomes clear that the rationale for exclusion rights thus actually reduces as the gap between the Cost of Creation and Cost of Replication reduces. And it is obviously the case that in several different spheres, GenAI tools are reducing the cost of creation for creators all around the world.
Thus, if we start from the foundation rather than the top, it is clear that doubling down on “IP” rights as we understand them, in fact, becomes less justified. Giovanni Sartori, in 1970, described conceptual stretching to reference situations wherein thinkers, in applying known or familiar frameworks to new conceptual terrain, end up transferring existing labels even while losing the analytical precision that gave the framework its raison d’etre to start with. And this is what we seem to be seeing here with the grasp towards the familiar, comfortable tool of IP, to solve problems that are at best only partially connected to them!
IP for the sake of IP was never the purpose. However, given the increasingly expansionist tendencies of IP policies over the last few decades, it is certainly what we need to watch out for. The dangers of remembering the reward but forgetting the reason for the reward are, in fact, well documented in economic policy (even if unclear on historical accuracy) as the pervese incentive, or the Cobra effect! In an ideal world, we would not only actively be trying to interpret current exclusionary rights in a manner more calibrated towards this reducing costs gap, but to also be able to restrict the scope, as and when this calibration requires it.
The dangers of remembering the reward but forgetting the reason for the reward are, in fact, well documented in economic policy (even if unclear on historical accuracy) as the pervese incentive, or the Cobra effect!
It also betrays a lack of imagination that even as we entrap ourselves into simply looking at old solutions to solve new problems, we are still somehow missing rather straightforward methods that could bring about revolutionary changes because of how we’re stuck in an IP frame of thinking. For example, if we truly wanted to ensure a robust creative corpus, imagine what a mandated public library cum community space, in every village in India, could do for the country? Not just the economy, but for people, for society, and for enriching our cultural and knowledge heritage!
Value vs. value
This brings us to the most important question: What is the problem we’re actually trying to address? Earlier thought on IP showed us that what IP tried to do is to ultimately ensure a flourishing public sphere or public domain, even if the method required temporary exclusions, to ensure creators are able to survive in a market economy in their process of contributing to the replenishment of the publics. If one teases this method apart, there is a fundamental flaw that is present, and that the GenAI world is ripping wide apart, due to its unprecedented extractive capacity. Let’s dig into this.
Right now, mirroring the familiar IP framework, much of the discourse in the GenAI context, when it cares about the well-being of non-AI actors, is framed around adequate compensation. However, this framing, inadvertently, reduces the value of culture (represented here through cultural artifacts) to mere transactional commodities. Our concerns, though, are not about ‘market value’. Rather, they are about the actual (de)value(ing) of culture, the public domain, and the public sphere, and more. This latter set of “values” refers to concepts which are constitutive of human flourishing, cultural and democratic participation, enhancement of social community and capability, and more.
Our concerns, though, are not about ‘market value’. Rather, they are about the actual (de)value(ing) of culture, the public domain, and the public sphere, and more.
In other words, though we are using the same word ‘value’, we are looking at two different concepts, which, for the purposes of this essay, I will refer to as ‘commodification value’ on the one hand, and ‘cultural or public value’ on the other. And thanks to the powerbrokers and governance-brokers of our times, where efficient commoditization gives superstar status, we as a people, have a fairly good understanding of commodification value, while very little articulation has gone into a theory of public value.
For example, it is fairly obvious to critics of centralization of power that the current AI ecosystem presents a great risk of centralization of extreme power, as the top AI companies are ‘extracting value’, while ‘value creators’ at the bottom are structurally not positioned to see much growth. (see more in an earlier blog post I’ve written here). However, even here, these are fundamentally different concepts of ‘value ‘. The former is focused on what can be commodified, whereas the latter encapsulates value that is so much more than that. This makes the problem far more than mere difference in accumulation. Further, reducing the latter to a commodified value reflects a market-value framework, which is de facto commodification, and thus devalues culture. Conversely, by virtue of ingrained market thinking, culture, the public domain, the commons, the public sphere, etc all remain undervalued because they can only be partially commodified, if at all. This constant tension between what we do knowingly value, and what we don’t articulate yet still value is one that needs to be addressed and reflected in appropriately articulated policy goals, and not shoe-horned into “IP” policy.
The reform agenda
While there may be several different ways to approach these questions, I’d like to indicatively highlight a couple of ideas from UCL’s Institute for Innovation and Public Purpose.
First – Mazzucato et al’s 2020 work, where they propose a threefold Public Value framework that looks far beyond what can be commoditized, combining concepts from three levels – individual value, industry value, and societal value.
And second, Mazzucato’s 2025 work synthesising ‘culture’ and ‘industrial strategy’, where a very useful governance template is proposed. If we wanted to focus on Cultural Commons, then borrowing the 5 pillars from there could, for example, lead to the following governance questions that require focused articulation:
- Purpose and directionality (What is the Cultural Commons for?)
- Co-creation and Participation (Who gets to shape its governance)
- Knowledge Sharing and Learning (How are insights from Cultural Commons use fed back)
- Benefit and Reward Sharing (How are gains distributed)
- Transparency and Accountability (What disclosure obligations attach?)
It is clear that to even start developing better articulations on centering a public domains and commons framing, knowledge governance mechanisms must be taken more seriously from interdisciplinary and poly-centric lenses, i.e., to more seriously engage with policy from the depths of different disciplines such as Competition, Labour, Tax or other Social policies, along with IP.
And finally, a serious knowledge governance architecture would also need to account for the political economy and how different stakeholders strategically utilise, as well as impact, the available tools of law and policy. Policy would accordingly need to not just be reactive, but to proactively account for temporal dynamics. These could include:
- Defensive Positioning to prevent further enclosure (transparency mandates, reduced scope of copyright/patent/trade secret claims, enforcement of relevant protections from other domains such as labour law, competition law, etc),
- Wide Redistribution to ensure gains from AI that is built from the public sphere/public domain/commons are shared more broadly (heavy cesses or taxation, direct support – funds and otherwise – for creators or knowledge workers, etc)
- And Intentional Generative policy to proactively build institutions that are purposed towards making the knowledge commons thrive (open data infrastructure, support for interoperable and open AI architectures, etc).
It’s clear that there will be no simple path forward. However, perhaps by starting to ask better questions of what we actually are looking for, and relying less on existing policy proxies and presumed meanings, we will eventually come to better answers!
The series brings together expert voices and was commissioned to inform the development of the issue brief by IT for Change, ‘Governing AI for the Cultural Commons: Beyond Intellectual Property’, under the AI, Culture and Intellectual Property Subgroup of the UNESCO Global Civil Society Organizations (CSO) and Academic Network on AI Ethics and Policy.
New article every Wednesday! Watch this space for more thinkpieces and read the issue brief here.
