The Unseen Risks of AI in the Offshore Energy Sector

Abstract Artificial intelligence (AI) introduces risks that existing offshore safety mechanisms are not designed to address. These risks are not easy to imagine major hazard risks, but less detectable systemic risks that can quietly lead to disaster. As humans and AI interactions become more integrated, they become more unpredictable. Emergent properties from these interactions can erode traditional safety barriers, limit the effectiveness of regulations, undermine operational processes, and compromise system safety design – thereby increasing the risk of systemic failure. This paper aims to contribute to the adoption of AI for safety-critical applications by proposing a safety case template for systemic AI safety. The template proposes a novel form of safety case, based on existing assurance elements, for a specific industrial application of AI. We hope this template can contribute to the safe use of AI in the offshore energy sector and stimulate discussion about novel risks introduced by AI systems.