Artificial Intelligence in Carbon Capture, Utilisation, and Storage 2025 Workshop

Overview

The IEAGHG AI in CCUS Workshop, held virtually on 29–30 April 2025, was convened to foster a deeper understanding of the role artificial intelligence (AI) can play in advancing carbon capture, utilisation and storage (CCUS). The first day, with 266 attendees, focused on fact-finding: reviewing the present status of AI applications across the CCUS value chain and showcasing case studies that demonstrate both the capabilities available today and the potential AI holds for the near future. The second day shifted towards interactive, collaborative discussions with 43 invited experts, aimed at unpacking the risks, opportunities, and barriers associated with integrating AI into CCUS operations. Through this process, the workshop sought to identify research gaps and recommendations for addressing them.

Key Findings

• AI is a transformative enabler for CCUS, i.e. has demonstrated significant potential to accelerate innovation, improve operational efficiency, and reduce costs.
• Environmental and ethical considerations must be addressed. The energy intensity of AI models, potential biases, and risks of overreliance were highlighted as concerns.
• Real-world applications are emerging, enhancing material discovery, automating subsurface analysis, streamlining permitting, and monitoring CO₂ transport and storage.
• Trust, transparency, and explainability are critical. Thus, applications require models that are interpretable and auditable. Black-box systems face resistance, especially in permitting and public engagement.
• Data remains a major bottleneck. High-quality, diverse, and standardised datasets are essential for training robust AI models. Lack of these will likely lead to increased hallucinations.
• AI should augment, not replace, human expertise. Across all sessions, participants agreed that AI must support human decision-making rather than automate it entirely.

Following on, we recommend the actions below to overcome the concerns identified above:

• Establishment of cross-sector data sharing frameworks between industry, academia, and government to unlock proprietary and siloed data.
• Investment in explainable and interpretable AI to ensure model transparency from the development phase.
• Development of benchmarking and validation protocols and alignment of AI validation with traditional engineering and scientific methods.
• Supporting interdisciplinary training and capacity building, i.e. training CCUS professionals in AI and vice versa.
• Accelerating permitting with AI tools to reduce review times and associated costs and improve completeness checks.
• Promotion of energy-efficient AI development by e.g. developing low-power models and investing in low-carbon energy-powered data centres.
• Embedding AI into digital twins and monitoring, reporting and verification (MRV) systems to enhance real-time monitoring, predictive maintenance and carbon credit issuance.
• Alignment of AI deployment with evolving policy frameworks, such as the EU AI Act and Article 6 of the Paris Agreement.