On 1 January 2024 The Petroleum Safety Authority Norway (PSA) became The Norwegin Ocean Industry Authority (Havtil).

“One of our jobs is to ensure that the regulations and supervisory methods are relevant and suitable in the encounter with new technology,” says Finn Carlsen, the PSA’s director of professional competence. “As a result, we’ve devoted particular attention in recent years to digitalisation and ICT safety.”

A significant part of this commitment has involved learning what is involved, mapping requirements and enhancing expertise about risk related to these issues. That has included looking at various dimensions and aspects of digitalisation and the robustness of industrial ICT systems.

“Our work on artificial intelligence (AI) builds further on the knowledge we’ve acquired and studies conducted in connection with our commitment to ICT safety,” Carlsen explains.

Photo of AI-generated platform offshore
AI generated photo (Midjourney) Photo: Havtil

Increased use of AI

Over time, the PSA has seen examples of AI being applied in the petroleum industry, and expects its use to increase in a number of areas over coming years.

“It’s in an early phase, with testing and product development,” Carlsen observes. “At the same time, we see that the industry wants to buy ready-made products which have often been developed for other sectors and purposes in order to reduce costs and simplify integration with existing digital ecosystems.

“Over the next three-five years, we’ll be devoting greater attention to AI-related safety in the petroleum industry. Our ambition is to establish favourable operating parameters for prudent AI use while actively following up and seeing to it that the industry manages risk factors associated with developing and maintaining AI solutions.”

In this context, the PSA wants to help identify and secure an overview of approaches for managing risk associated with AI by:

  • helping to improve guidance on and understanding of regulatory requirements, and how these contribute to safe development and use of technology
  • strengthening the PSA’s enforcement powers (regulation, supervision and advice) related to following up the AI industry.

“Following up the risk factors associated with AI calls for a cross-disciplinary approach and specific methods for identification and management,” says PSA project manager Linn Iren Vestly Bergh.

“Our attention is primarily concentrated on prudent use of AI in the petroleum sector as a high-risk industry with a major accident potential.”

Defining AI

Many different definitions of AI exist, and these change in line with technological developments. The PSA has no definition of its own, but applies the one utilised in Norway’s national strategy for AI and the Norwegian Digitalisation Agency’s web pages on this subject.

The national strategy defines AI as “systems which perform actions, physically or digitally, based on the interpretation and processing of structured or unstructured data in order to achieve a defined goal”. AI includes various approaches and techniques, such as machine learning.

Regulations

Norway’s regulations on health, safety and the environment in the petroleum sector are performance-based, technologically neutral and built on risk management, Bergh notes.

“We therefore believe they’re already relatively well-suited as they currently stand for following up AI solutions. But they lack references to norms and standards which can provide guidance on their use.

“We’re keeping an eye on standardisation efforts by the industry, and following this up in our work on the regulations.”

Risk

Introducing AI solutions will lead to changes in duties, work processes and collaboration. That has a particular impact on the risk picture when a gap opens up between technology and work processes.

Where the petroleum sector is concerned, solutions which might affect the possibility of major accidents or pose threats to safety systems, barriers or critical infrastructure could be regarded as high-risk applications of AI.

Using AI models for operations, planning and decision support could also directly or indirectly affect safety, the working environment, emergency preparedness and security in petroleum activities.

“An integrated risk understanding is the foundation of good safety – and that also applies to AI,” Bergh emphasises.

“Risk-reduction principles, such as clarity,  transparency and verifiability, are therefore important for safety when AI is developed and adopted in the petroleum sector.”

Networking

Collaboration with selected Norwegian centres of AI expertise will form part of the PSA’s future work in this area. Active collaboration with academia and other industry players in the sector is pursued to concentrate attention on safety-related risks in the development and use of AI for petroleum operations.

Innovation Day 2024

As part of this initiative, the Innovation Day seminar in 2024 will address responsible use of AI from a safety perspective. This event is an important meeting place where the whole industry can exchange experience and learn from each other.