Accountability Gaps in AI-Enabled Climate Decision-Support Systems

Artificial intelligence–enabled decision-support systems are now routinely used in climate policy, planning, and international reporting, shaping emissions estimates, mitigation pathways, adaptation priorities, and transparency assessments under the Paris Agreement. Despite this growing influence, these systems continue to be governed as technical tools rather than as institutional actors, creating accountability deficits that climate governance frameworks are not equipped to manage. This paper takes the position that AI-enabled climate decision-support systems function as de facto governance infrastructures whose outputs carry political, distributive, and accountability consequences. Treating them as neutral analytical aids obscures how algorithmic design choices encode assumptions, prioritize policy options, and reallocate epistemic authority within climate regimes. As a result, existing transparency practices, centered on disclosure of outputs and methodologies, are insufficient to sustain accountability and institutional credibility.

The paper reframes transparency in AI-enabled climate governance as an enforceable institutional condition grounded in explainability, traceability, and contestability. It identifies a structural governance gap arising from the deployment of algorithmic systems without explicit mandates, responsibility allocation, or alignment with climate reporting architectures. Left unaddressed, this gap risks undermining national reporting credibility, distorting adaptation planning, and weakening trust in collective climate processes. The analysis establishes governance, rather than innovation, as the decisive factor shaping the legitimacy of artificial intelligence in climate decision-making and delineates the minimum institutional conditions under which AI-enabled systems can be integrated without eroding the foundations of climate transparency and accountability.