The UK’s Office for Nuclear Regulation (ONR), the US Nuclear Regulatory Commission (NRC) and the Canadian Nuclear Safety Commission (CNSC) have jointly published a paper, Considerations For Developing Artificial Intelligence Systems In Nuclear Applications, looking at the principles that all participants in the AI lifecycle should consider, including developers, licensees and regulators in the nuclear sector.

The 27-page paper covers topics such as: common areas of focus for regulators when approaching AI; understanding how to manage systems based on the consequences of AI failure and the level of AI autonomy; the importance of human and organisational factors in the use of AI; integrating AI into existing nuclear systems; managing the AI lifecycle from design to deployment; and considerations for safety cases for AI in nuclear applications.

ONR Technical Director Shane Turner said: “This significant collaboration between CSNC, NRC and ONR will support the wider international nuclear community to understand what is important when considering the application of AI. ONR is open to innovation and is committed to enabling the safe and secure deployment of AI and other innovative technologies within the nuclear sector.”

The guidelines for secure AI system development were formulated in cooperation with 22 agencies and ministries from across the world, including security agencies from the UK, US, and Canada. The document recommends guidelines for any systems that use AI, whether those systems have been created from scratch or built on top of tools and services provided by others.

The document is split into several sections that the Canadian, UK, and US nuclear regulators consider may be important in managing the risks arising from the use of AI. These include:

  • Use of existing safety and security engineering principles;
  • Human and organisational factors;
  • AI architecture;
  • AI lifecycle management; and
  • Documenting AI safety and security.

The paper concludes: “The nuclear industry benefits from decades of operational experience, mature and rigorous design and operation protocols, and a strong safety and security culture. The rapid pace of recent AI development is somewhat antithetical to the slow and methodical change process that the nuclear industry traditionally follows. Nevertheless, the primary goal for the nuclear industry and regulators with respect to AI systems will be maintaining adequate safety and security while benefiting from their deployment.”

Furthermore: “The fast pace of AI development means it is unlikely that AI-specific consensus standards for the nuclear domain will be available to support regulatory activities within the near future. In the interim, existing nuclear-specific standards remain a starting point coupled with considering the unique attributes introduced by AI.”

Finally: “International standards exist for conventional software, hardware, and systems development in the nuclear sector. However, standards for nuclear-specific AI design and substantiation do not currently provide sufficient assurance in high consequence applications or where societal harms might exist. The fast pace of AI development means it is unlikely that AI-specific consensus standards for the nuclear domain will be available to support regulatory activities within the near future. In the interim, existing nuclear-specific standards remain a starting point coupled with considering the unique attributes introduced by AI.”