Navigating the frontier of data openness: the obligation to cooperate in marine climate data governance under the AI Era

Artificial intelligence (AI) models are increasingly used in marine climate monitoring, prediction, and decision support, yet their reliance on large-scale training data has exposed a structural imbalance in marine climate data governance. Data-contributing States, especially Small Island Developing States (SIDS), may provide critical local observational data while lacking access to early-warning products, localized decision-support tools, and model capabilities commensurate with their climate vulnerability. This article reassesses the obligations of marine environmental protection, cooperation, information exchange, and technical assistance under the United Nations Convention on the Law of the Sea (UNCLOS) in the context of AI-enabled marine climate governance. Through doctrinal legal analysis and evolutive treaty interpretation, it examines how UNCLOS can respond to the transformation of marine climate data from shareable scientific information into AI-derived model capability. The analysis identifies three interrelated dilemmas: the decoupling of data contribution from model benefits, the market-based restriction of predictive services needed for public-risk governance, and the erosion of trust caused by opaque downstream data use and dual-use risks. It argues that conditional data openness should be understood as an interpretive specification of the duty to cooperate under UNCLOS in the AI era. This framework combines purpose limitation, procedural transparency, and fair reciprocity to ensure that data openness remains linked to public marine climate risk governance, traceable oversight, and model-capability-oriented technical assistance. This approach offers a legally grounded pathway for aligning AI-enabled marine climate governance with cooperation, equity, and the protection and preservation of the marine environment under UNCLOS.