Applying Common Heritage of Humanity Principles to Artificial Intelligence
By Matthew Hoisington
***
OpenAI recently made an extraordinary announcement. It would place five percent of the value of its company in public hands through a mechanism modeled on the Alaska Permanent Fund, which provides annual dividends to its residents based on petroleum resources. OpenAI acted alone, but its proposal envisions that other artificial intelligence (AI) developers and companies, such as Anthropic and Google DeepMind, would contribute similar percentages to the public.
The proposal marks a significant shift in thinking. Instead of private companies and shareholders enjoying massive windfalls from AI’s development, it would provide the public with vested financial interest. It also recognizes something that might seem obvious: AI is made possible by all of humankind, and we should all benefit from it, financially and otherwise.
International law has a name for this principle, called the common heritage of humanity. The idea emerged in 1967, when Arvid Pardo, a diplomat from Malta, one of the smallest and least powerful states in the United Nations, proposed in the General Assembly that the deep seabed and its resources should be managed “for the exclusive benefit of mankind as a whole”. Pardo’s formulation had four elements: the ocean floor would not be subject to national appropriation; it would be managed internationally; its benefits would be shared equitably among all peoples; and it would be used for peaceful purposes and preserved for future generations.
The principle shaped later agreements on the law of the sea, the law of outer space, the Moon Treaty, and international environmental law. It is a principle that speaks for all humanity, recognizing our shared interests and common endeavors. While its record of implementation has been uneven and subject to contestation, the principle has endured. Some things are too consequential, and too collectively derived, to be controlled by any one nation or private enterprise. They belong to us all and should be managed accordingly.
The argument for the inclusion of AI in the common heritage of humanity principle is strong. Unlike the ocean floor and celestial bodies, AI is engineered by humans, often through a mix of public and private capital. But these systems are distilled from, and rely upon, the accumulated knowledge of humankind. They draw on centuries of literary and scientific achievement, the open web written by millions of people in hundreds of languages, open-source code contributed freely by generations of programmers, and research financed by taxpayers. AI continues to learn every day from the input of hundreds of millions of users. The engineering is largely private, but in essence, the raw material is our collective knowledge.
If AI is viewed through this lens, then a common heritage of humanity framework comes into focus. If embraced and applied, it provides a blueprint that can help organize an uncertain future.
The architecture that Pardo and his successors designed for the deep seabed, which foresees a common resource that generates royalties distributed equitably to humanity, is precisely what OpenAI has turned to. Proposals by politicians in Washington range from five percent of the leading AI companies to far more ambitious public shareholdings. None of the proponents cite the common heritage of humanity principle, but they have arrived at the same design.
International management is also progressing. In the past two years, UN member states have adopted the Global Digital Compact and established an Independent International Scientific Panel on AI and a Global Dialogue on AI Governance. These are early-stage institutions, and it is easy to dismiss them as deliberative and lacking enforcement capabilities, but they are recognizable as the early scaffolding of collective stewardship for AI. The International Seabed Authority plays a similar role for the ocean floor.
Peaceful use will not be easy given AI’s breadth and applicability in military contexts, but the growing debate over autonomous weapons and the role of human operators in warfare demonstrates an attempt to regulate AI’s potential for negative impact. The idea that today’s AI developers have a duty to future generations to ensure AI’s safe adoption represents another thread in this line of thinking.
None of these efforts have set out to build a common heritage framework for AI. They are fragmented and lack coherence. But the common heritage of mankind regime is emerging despite this fragmentation because the principles it enshrines could help shape how it is being developed, made available and regulated.
Applying these principles to AI will attract criticism, including that it could stifle free markets and innovation. These are misplaced. The international seabed regime licenses commercial mining and returns profits to investors. Benefit-sharing and private enterprise have coexisted for decades. It is also worth noting that the five percent proposal for AI originated from OpenAI, a titan of the industry itself. The innovators evidently do not regard a public stake as detrimental to their business model.
The principle may also be criticized as naïve in an era of great power competition and transactional politics, but this ignores the historical context in which the principle emerged. The common heritage of humanity framework was born at the height of the Cold War, championed by a small state, precisely because rivals will sometimes prefer neutral rules applicable to all as opposed to an unwinnable race for control.
The key to success for a new AI principle is to make its parameters actionable. This will require compromise and negotiation, but it is worth the effort. Taking current regulatory efforts on AI and its potential for development and economic impact (both positive and negative) into account within a coherent frame connects scattered initiatives with a moral core. It also provides a means to ask, at each successive step, whether regulation is working and whether, as more is known about the technology’s future, it remains an appropriate, up-to-date framework. Adjustments and specializations are inherent in common heritage of humanity applications as they are consistently revisited to account for the economic realities and unique considerations of their subjects. The analysis and institutional arrangements can remain agile and responsive.
The UN remains the best platform to define, advocate, and enforce the principle. A General Assembly resolution that recognizes AI as a common heritage of humanity would be non-binding, but that could be one of its strengths. Such an instrument will only work because actors come to believe in its merits. It would cost nothing and orient everything, just as Pardo’s speech did for the oceans.
Governments and institutions were slow to respond to Pardo’s declaration, but they increasingly came to see the benefits of such an approach. The translation of the principle to a working structure has admittedly not been easy. Renegotiations have been necessary, and disputes have arisen. The same will occur for AI, particularly given its rapid advancement and uncertain future. But these discussions will take place within a shared language and set of objectives that the principle provides.
Pardo spoke on behalf of a nation of a few hundred thousand people and helped to rewrite the law governing half the planet’s surface. He named an evident, mutually beneficial truth before the powerful had settled the question. The truth of our moment is that AI was built from our collective knowledge and history. OpenAI’s announcement shows that even its builders sense the dividend should be shared. The practice is arriving ahead of the framing. Supplying the frame is how a gesture becomes a commitment, and how a technology made from everything we have ever known becomes, in law as in fact, the common heritage of humanity from which we can all benefit.
Matthew Hoisington is currently the Legal Adviser for the Gang Suppression Force in Haiti authorized by Security Council resolution 2793 (2025). He has 15 years of experience in the United Nations system with the United Nations, Office of Legal Affairs; UNICEF; WFP; UNRWA; and two United Nations peacekeeping operations in Darfur and Abyei. He is a graduate of the Fletcher School (LLM, 2012).
