Strengthening and safeguarding artificial intelligence for a planet under pressure

Abstract Artificial intelligence is rapidly reshaping sustainability science through advances in climate modeling, ecosystem monitoring, resource management, and evidence synthesis. Yet its use also raises concerns about reliability, equity, and environmental impacts. In this Prospective, we identify four priorities for responsible AI integration: ensuring AI-supported claims are causal, transparent, and auditable; treating model selection as an environmental decision; building sustainability-focused AI infrastructure; and engaging sustainability scientists and affected communities in AI design and governance. With these safeguards, AI can enhance sustainability science while promoting equitable and environmentally responsible decision-making.