The research programme

Marine genetic resources (MGRs) underpin a multibillion-dollar pharmaceutical and biotechnology sector, yet the international legal architecture governing who benefits from this value was designed for a world of physical collection and traceable biological samples. Artificial intelligence changes this. Researchers can now use machine learning models to predict molecular structures, screen bioactive compounds and identify commercially valuable discoveries computationally, often without ever handling biological material. When discovery no longer requires physical access, the legal rules designed to govern that access may no longer apply.

This research programme investigates what happens to international environmental law when the activity it regulates becomes invisible to its own instruments. The programme spans doctrinal analysis of treaty law, structured legal knowledge representation and the design of AI-assisted tools for legal information retrieval.

How AI dissolves three foundational legal concepts

The governance architecture for marine genetic resources rests on three foundational concepts: access, origin and benefit-sharing triggers. Each presupposes that the regulated activity can be defined with sufficient precision to activate the relevant legal obligation. AI-enabled bioprospecting dissolves all three.

Access. Under the Nagoya Protocol and the BBNJ Agreement, benefit-sharing obligations are triggered by ‘access’ to genetic resources. But predictive AI models aggregate thousands of sequences of mixed and often unknown provenance to generate novel compounds computationally. There is no discrete sequence being accessed, no identifiable organism being utilised, and no single jurisdiction from which permission must be sought.

Origin. The WIPO Treaty on Genetic Resources (GRATK) requires patent applicants to disclose the ‘country of origin’ of genetic resources. EU Regulation 511/2014 requires due diligence on the provenance of genetic material. But when outputs emerge from patterns across entire datasets rather than from any single genetic resource, provenance becomes structurally untraceable.

Benefit-sharing triggers. When neither access nor origin can be reliably identified, the benefit-sharing architecture loses its operative foundation. The distributional consequences fall hardest on the biodiversity-rich regions and communities the regime was designed to protect.

This programme argues that these concepts dissolve not because AI is too new for the law, but because the concepts were designed as if they had essential definitions. Drawing on Wittgenstein’s notion of family resemblance, the research shows that physical collection, digital download and computational prediction all share some features of ‘access,’ but no single feature is common to all three. They form a family: each resembles the others in certain respects, but there is no common essence that unites them. Governance mechanisms built on definitional precision will systematically fail whenever a new mode of utilisation enters the family.

Mapping legal positions across treaties

The programme examines how six international instruments interact, and where they fail, when bioprospecting becomes computational: BBNJ Agreement (2023) · Convention on Biological Diversity (1992) · Nagoya Protocol (2010) · TRIPS Agreement (1994) · WIPO GRATK Treaty (2024) · UNCLOS (1982).

Using Hohfeld’s analytical framework, the project maps the legal positions (claim-rights, duties, powers, immunities) created by each instrument. The resulting mapping reveals where obligations overlap, where they conflict and where AI-driven bioprospecting falls through the gaps. The analysis exposes persistent asymmetries: UNCLOS protects freedoms on the high seas but disables collective entitlements; the CBD and Nagoya Protocol grant provider states powers that stop at national borders; TRIPS entrenches proprietary entitlements without requiring origin disclosure; and Indigenous Peoples and Local Communities remain structurally marginalised, holding no enforceable rights over resources beyond state borders.

A core output of the programme is a manually curated, machine-readable dataset that serves as a resource for further research.

Trustworthy AI for legal information retrieval

The complexity of this fragmented legal landscape falls hardest on actors from the Global South: Small Island Developing States, Indigenous Peoples and Local Communities, and under-resourced institutions that lack the capacity to navigate thousands of pages across six instruments. Artificial intelligence could help democratise access to this legal information, yet opaque large language models risk reproducing the very power asymmetries they promise to address.

To meet this challenge, the programme develops a Hohfeld-Structured Normative Knowledge Base (HSNKB) that translates treaty provisions into explicit legal positions, making visible who owes what to whom, under which conditions and where normative gaps persist. Combined with LLM-assisted retrieval, the knowledge base keeps every result traceable to specific treaty text while preserving competing interpretations. The aim is to redistribute interpretive capacity toward marginalised actors.

Governance design

The interpretive choices currently being made by Conferences of the Parties, patent offices and EU regulators will determine whether AI-enabled bioprospecting falls within or outside the benefit-sharing regime. The programme examines emerging governance responses, including the Cali Fund’s sector-based model for benefit-sharing from digital sequence information, the GRATK’s disclosure obligations and their enforcement gaps, and EU Regulation 511/2014’s due diligence framework. The research proposes that mechanisms of benefit-sharing should be organised around family resemblance rather than definitional precision, extending the Cali Fund’s logic beyond its current scope to build governance that is resilient to technological change.

Selected publications

Marcos, H., Nanda, R. & Schütz Veiga, J. (2026), ‘Marine Genetic Resources, Who Owns and Who Owes What? A Hohfeldian Mapping of Legal Positions on MGRs’ — Leiden Journal of International Law.

Deep Ocean Stewardship Initiative (2025), ‘Towards Coherence and Avoiding Undermining: Policy Recommendations on Implementation of the BBNJ Agreement Regarding Marine Genetic Resources’ — Policy Brief.

Nanda, R., Marcos, H., Westermann, H. & Schütz Veiga, J. (2025), ‘A Hohfeldian Knowledge Base for LLM-Assisted Legal Information Retrieval in Marine Biodiversity Law’ — In: Markovich, R. et al. (Eds.), Legal Knowledge and Information Systems (Vol. 416, pp. 318–323, IOS Press).