Following a joint technology declaration by NHO and key Norwegian industrial companies in January, DNV led the kick-off of the workstream to operationalize a regulatory sandbox for high-risk industrial AI, bringing leading Norwegian companies together to turn intention into implementation. On 19 March 2026, DNV hosted the first working session in a national effort to deploy artificial intelligence safely and effectively in industrial settings. The participating companies presented the industrial AI use cases they see as most critical in their sectors, including autonomous navigation, fraud detection, process optimization, and predictive maintenance. These use cases highlighted the consortium's diversity: telecommunications, finance, energy producers, heavy industry, maritime and defence technology providers, and advanced manufacturing companies all brought fundamentally different operational contexts and risk profiles to the table. Yet despite this diversity, the group quickly recognized the need to identify shared challenges that cut across sectors. Several participants emphasized that one of the most important early tasks for the sandbox will be to define which industrial AI applications qualify as high-risk, and what system-level requirements must be in place to ensure their safe use. Participants also noted that the consortium is working in what was described as a 'blue ocean': there is no pre-existing model for an industrial, cross-sector AI sandbox of this kind. This gives the group a unique opportunity to shape the sandbox to promote shared objectives. A recurring theme was the need to scale autonomy safely. Many of the use cases discussed involve systems that influence or control industrial assets and processes – vessels, power production plants, and industrial machinery. These systems bring distinct safety, security, and trust requirements, and require rigorous verification and validation methods such as simulation, scenario exploration, uncertainty analysis, and claims testing. Participants also pointed out that delivering this safely and effectively will require both deep domain expertise and advanced algorithmic expertise. Domain specialists understand operating environments, hazards, and engineering constraints, whereas AI specialists understand model behaviour, robustness, explainability, and data-driven risks. Only by combining these perspectives can the group define meaningful requirements and credible assurance methods. The workshop also surfaced foundational questions that must be addressed early: How do we define 'AI' for the purpose of the sandbox, given the wide range of algorithmic systems used across industries? How will frontline personnel interact with these systems, and what are the implications for human trust, workload, and expertise over time? What evidence is needed to demonstrate safety and performance, and how should that evidence be generated, structured, and communicated? The kick-off concluded with a shared commitment to develop a sandbox that advances safety and innovation across industries. If successful, the consortium's work could provide Norway with a first-of-its-kind framework for responsible, scalable, and innovation-enabling industrial AI, strengthening its position at the forefront of safe digital transformation. About the national initiative NHO's declaration brings together leading companies to strengthen Norway's competitiveness in industrial AI and quantum technology. The signatories are Aker, Cognite, DNB, DNV, Equinor, Hydro, KONGSBERG, SINTEF, Telenor, Vår Energi, and Yara.