The survey is part of Havtil's efforts to monitor AI in the petroleum industry. The 2026 sector report builds on the guidance letter concerning AI and safety that Havtil sent to the industry in May 2025.

Varying levels of maturity among the companies

The survey reveals differences in how operating companies manage and monitor AI systems that could have an impact on safety. Some companies have established registers, roles and processes to identify, classify and monitor the systems. Others have a limited overview of the AI systems that are under development, being procured or in use.

The differences relate for example to how companies describe their AI systems, the applications they consider to be relevant to safety, and how AI-relatedrisks are incorporated into management, requirements, documentation and follow-up.

Some companies refer to AI primarily as a digital tool or decision-making aid. Others view AI as part of a broader interplay between human, technology and organisation. Thus,  the technology can impact work processes, priorities, decisions and barriers.

An overview is essential for effective risk management

An up-to-date overview of AI systems is a prerequisite for risk-based monitoring. Many companies lack a comprehensive overview of the systems that are being procured, under development or in use. This also applies to AI that is incorporated into products and services from suppliers.

Such overviews make it possible to identify systems that could impact people, operations or safety-critical decisions. It thus provides an essential basis for the companies' risk management.

Prudent development and use

The analysis of the companies' responses raises questions about how the companies understand HSE regulations and how they assess whether AI systems could have an impact on safety.

Several companies reported that they do not have AI systems that could have an impact on safety. However, some of those same companies also describe the use of AI in well planning, maintenance management and the analysis of data to support operational decision-making. This may indicate that AI used as a decision-making tool is not always considered to have an impact on safety, even though such systems can influence decisions and priorities that could affect safety.

According to the HSE regulations, companies must identify and address both direct and indirect risks associated with the use of AI. AI systems can have a direct impact on safety if a model's results are used in the control or regulation of operational processes. AI can also have an indirect impact on safety, even if the system does not directly control processes. Results from AI can be used as a basis for maintenance decisions, the planning of drilling operations, condition monitoring or the assessment of non-conformities. Systems that “simply” provide recommendations, analyses or a basis for decision-making can thus affect safety through how information is interpreted and which measures are prioritised.

Further follow-up

The survey shows that, although the industry has begun to manage and monitor AI, there is still a need for a more systematic approach. Havtil will use this knowledge in further dialogue, guidance and supervision of the development and use of AI in the petroleum industry.