ID: 2243
Presenting Author: Silvia Fregoni
Session: 686 - Climate Change Impact Assessment: Countering Misinformation with Science
Status: pending
The paper shows how invisible AI emissions distort climate governance and argues that stronger disclosure rules can support more accurate impact assessment and sustainable AI decisions.
Current narratives around artificial intelligence fuel misinformation in climate governance, overstating its environmental benefits while obscuring its carbon costs. While AI supports climate action by optimizing energy systems and forecasting environmental change, it also harms the planet through its growing emissions—emissions that remain invisible to law and society. Understanding these hidden impacts is critical for environmental and social impact assessments, helping to develop mitigation strategies and guide decision-making toward environmentally sustainable AI operations.
This paper examines how climate-disclosure regimes—under the ISSB, CSRD, and California laws—and emerging AI regulations sustain that invisibility. Through comparative analysis of legal frameworks and corporate disclosures, it shows that AI-related emissions are buried within broad categories such as “data centers” or “information and communication technology,” which obscure AI’s contribution to global emissions and impede science-based assessment.
The paper argues that this invisibility is a socio-legal construction: disclosure rules define what counts as harm, and institutional fragmentation dictates who must disclose. Strengthening climate-disclosure regimes to make AI emissions visible would enable more accurate project assessment and evidence-based decisions about AI development. Rather than creating new regulators, the paper calls for recalibrating existing frameworks to ensure technological progress advances, rather than undermines, sustainability goals.
Silvia Fregoni, Senior Academic Fellow at Berkeley Law’s Center for Law and Business, researches ESG, climate disclosure, and AI governance in comparative regulatory systems.