Glenn Regge, Endeavor Management
December 2023 Offshore The magazine article, “Artificial Intelligence Applications Promises to Improve Drill Efficiency,” examined considerations for the upstream oil and gas sector using artificial intelligence (AI) for a variety of applications, from drilling optimization to completion design.
The use of AI usage by the upstream sector expanded very quickly, and 18 months later, Offshore We conducted research from our readers to determine the most important AI challenges and technologies facing the industry. As a result of the investigation, an article was published in June 2025. “The research highlights the biggest challenges in implementing AI technology in the offshore oil and gas sector.”
The study included operators, service companies and EPC companies, showing how AI usage is allocated in this market.
- 43% for predictive maintenance
- 31% of earthquake data interpretation
- 17% of drilling optimization, and
- 10% of reservoir management.
The research revealed that the industry's main concerns regarding AI use are:
- Integrate AI with existing systems
- Lack of skilled labor
- Data security concerns, and
- High cost.
This recent study also asked industry participants how AI enhances safety protocols in offshore operations. Includes:
- Computer vision for monitoring equipment
- Real-time data analysis
- Automatic inspection system
- Predictive risk assessment.
Following these analysis of industry baseline data on increasing use of generated AI, upstream sectors should decide how to:
- Determine the AI application and functional scope (SOW) for offshore construction based on EPC/EPCM contracts, seismic interpretation/reservoir management, facility operations, predictive maintenance and decommissioning.
- Addresses recognized benefits, risk/exposure, and specific applications, such as offshore construction, facility operations, predictive maintenance, and decommissioning, generally associated with AI sows.
- Contractually assigns use of AI and related exposures, including internal and external challenges (training, data collection/consistency, AI hallucinations).
- Procuring property, liability, and business interruption insurance coverage to address exposures arising from AI applications.
- Follow the rapidly evolving regulations and legal developments regarding AI use.
Some members of the offshore construction industry identify AI as a suitable tool for a variety of applications, ranging from supply chains, material procurement, project design, management/management to identifying failed installations. Some of the “real world” challenges associated with the use of AI for construction projects involve:
- Trainers with experience in these types of complex construction projects will use AI in a suitable and cost-effective way.
- “Education” AI on complex construction projects to address all elements/stages that may include all stages of the program, including engineering, procurement, construction, installation, commissioning, etc.
- Manage and use AI training and use, and create effective protocols that recognize/record situations where the AI generated may be affected by impaired data or hallucinations.
- Initial use of AI in a project may include inevitable financial and time consequences for project management due to concerns of “first use” that could affect project cost-effectiveness and scheduling.
- Properly resolve “conflicts” between generation AI management recommendations and existing HSE and regulatory requirements.
The challenges of these types of “first time” AI applications must be anticipated and addressed in construction contracts in a proper way.
EPC Contracts and AI
In January 2025, Westwood Global Energy announced that its 2024 EPC construction activities for Offshore Oil and Gas and Co.2 The Quarantine Project Award was $52 billion. Most offshore construction projects utilize existing types of contracts, such as engineering, procurement and construction management (EPCM). Engineering, Construction, Installation, Construction (EPIC); Engineering, Procurement, Construction, Installation, Commissioning (EPCIC) format. Westwood's analysis shows that the amount of EPC related to offshore oil and gas projects in 2025 could be in the $5.4 billion range.
Traditionally, under EPC contracts, contractors provide services for design, engineering, construction and procurement in a turnkey format, and provide approved subcontractors. Additionally, EPC contractors frequently assume responsibility for scheduling, delays and cost overruns.
Although EPCM agreements may have similar terms, EPCM owners often retain important construction risks, such as scheduling and cost overruns. Rather than working in the project manager role, EPCM contractors often act as advisory roles for owners.
Mintmesh, a US-based company founded in 2015, develops cloud-based programs for “engineers and unresolved engineering issues, including engineering, procurement and construction (EPC) projects.” Mintmesh references Bechtel as an example of companies using AI to evaluate project data, address scheduling issues, and use AI to “improve predictive capabilities.”
Sows with potentially generation AI in construction projects are evolving rapidly. A 2025 report from Autodesk, a global software company that creates technology for the architecture, design and construction industries, found that 76% of major construction companies are increasing their investment in AI. Construction projects using AI applications may contain some of the most variability and challenges in terms of contractual risk allocation.
Contractual risk allocation
The scope of AI applications and utilities for offshore EPC and EPCM construction projects is far beyond the scope of this article. However, the basic contractual risk allocation formats that have been used for years in these types of construction projects may need to be changed to address the risks arising from the use of generated AI.
“The Single Source of Truth”
When using generated AI in a construction project, parties must clearly state how AI will provide data from a “single source of truth” or “SSOT.” SSOT is the substantial source of data, information and regulatory guidelines for project SOW and specific AI applications. Additionally, parties must identify the means by which AI is “trained.”
The specification and use of SSOT is essential for generative AI to understand and incorporate a wide range of critical information regarding task assignments, agreed industry and regulatory standards/requirements, component capabilities, and manufacturing data. The SSOT designation requirement is a substantial component of the construction contract.
In the event of an AI-related construction defect, failure, or malfunction, specification of SSOT designation and compliance can be a key factor in determining the cause of or contribution to the failure. Furthermore, the claim may argue that the AI tool was not properly educated or trained by the manufacturer or user. AI manufacturers are expected to claim that the tool was misused.
In some AI-related construction disputes, it is expected that EPC or EPCM contractors or owners can pursue contractual claims against the AI manufacturer/creator. In some cases, an agreement between the EPC/EPCM contractor or owner and the AI maker/creator may limit liability for claims against the AI maker/creator.
AI monitoring
However, since generative AI tools need to rely on SSOT data, it is possible that the data may be exposed based on the conditions of the data, or that there may be data based on data negligence.
Parties using AI generated in EPC/EPCM contracts must contractually designate individuals to oversee AI work. It also requires a protocol for quality assurance and notification of issues to designated parties.
Risk allocation criteria
In many offshore construction contracts, parties rely on “mutual” or “knock for knock” risk allocation terms. These terms essentially require that each party be liable for the injury to its staff or damages to its property, as long as it arises from negligence rather than negligence by a third party. This “knock for knock” contract form is that each party is likely to insure for injuries to personnel or damage to their property.
Insurance coverage
Similar to the insurance industry's response to business-related claims arising from cyberattacks, the insurance industry is deciding whether and how to provide compensation for the damages arising from the use of generated AI.
In April 2025, Lloyd's Almira and Chaucer Insurance in London announced insurance coverage for the “risk of mechanical misperformance in AI systems and related liabilities.”
The announcement did not provide a detailed list of reports, but policy coverage includes “hapticism (false or misleading output), model drift (deteriorating performance over time), mechanical failures, and other deviations. Armilla/Chaucer coverage was created in collaboration with Lloyd's Lab Innovation Accelerator.
Testudo is another London 'startup' that provides data and technology for underwriting AI risk in the London market, backed by investments from Goldman Sachs. Testudo is also part of Lloyd's lab and plans to implement an AI coverage policy by the second half of 2025.
New Regulatory Framework
There were at least two important provisions regarding the use of generator AI in the recently passed “One Big Beautiful Bill Act.” OBBBA offers:
- Federal funding and tax incentives for companies investing in US-based AI development and infrastructure, including AI research, data centers, and semiconductor production.
- Strengthening restrictions on foreign influence and investment by “prohibited foreign companies” in AI activities supported by federal funds.
Presidential Order
There have also been many AI and data-related executive orders (EOs) from the Trump administration.
- EO 14318 Accelerate Data Center Infrastructure Permissions – To streamline the development of US AI-related infrastructure, this EO is aimed at data centers and infrastructure development. The US Department of the Interior and the Department of Energy will work together to approve appropriate federal land for energy development.
- EO 14320 Promotes exports of the US AI Technology Stack – This EO focuses on exporting US-based AI technology to reduce international dependence on AI technologies developed by AI enemies.
- EO 14319 Preventing awakening in the federal government – This AI directs the administration to issue guidance to federal agencies to implement the principles of fair AI.
The US private sector and federal government are investing enormous amounts of capital and resources in developing AI for large-scale commercial applications. Many of these applications include offshore energy development and large-scale capital projects.
The rapidly evolving commercial and regulatory AI development overview shows that industries are moving faster than regulatory frameworks aimed at managing AI development. Similarly, the insurance and risk allocation sectors are trying to analyze and “ring” the possible negative effects of AI that are exploited in untested or irrational ways. In Endeavor Management, this same phenomenon has been played in Deepwater Oil and Gas Exploration, where federal regulators have tried to quickly create regulations to analyze new technologies and manage their use. Hopefully, this evolutionary process will continue in a productive and reasonably safe way.
