AI in the energy industry: seizing opportunities, taking responsibility

What the EU AI Act will change from August 2025

The German energy industry is under enormous pressure to innovate: climate targets, volatile energy markets and the expansion of renewable energies require new approaches to planning, control and operation. Artificial intelligence (AI) offers considerable potential here: it helps with load forecasting, can control decentralised energy sources, enables optimised maintenance and drives the digitalisation of the grid infrastructure. But with great power comes great responsibility: the EU AI Act, which has been in force in its essential parts since August 2025, introduces a binding regulatory framework for AI applications in Europe for the first time. What does this mean in concrete terms for the energy industry? And why does AI remain an indispensable driver of innovation despite regulation?

Following the first step in February 2025, which banned certain AI applications with unacceptable risks, the next stage of the EU AI Act came into force on 2 August 2025. Since then, initial obligations have applied to general-purpose artificial intelligence (GPAI) models – from text generation to image creation. Transparency and documentation requirements apply in particular, with further provisions on the transparency and governance of these models coming into force in stages. For you as an energy company, this means it is high time to develop your AI strategy and adapt it to current and future conditions.

The EU AI Act – a milestone for trustworthy AI

With the AI Act, the European Union is creating a uniform set of rules for the development, provision and use of AI systems. The aim is to promote innovation while minimising risks to safety, fundamental rights and the environment. So-called high-risk systems, for example in grid operation, are particularly relevant for the energy industry. These are subject to stricter requirements on a staggered basis: high-risk systems will be regulated 24 months after the Act comes into force, with longer transition periods applying to certain product categories. Providers of GPAI models have had to comply with transparency and documentation requirements since August 2025, which also benefits energy companies if they use such models in chatbots for customer service, for example.

New realities, clear opportunities

The risk-based approach of the AI Act brings structure to the AI landscape and already provides clarity about upcoming regulations for future AI use cases. For the energy industry, which is traditionally highly regulated, this provides a familiar basis for designing AI projects to be legally compliant from the outset. Companies that act now will gain decisive competitive advantages: while others are still hesitating, they can already build their AI applications in a future-proof manner today.

AI use cases in focus: where opportunities and hurdles lie

Customer service: transparency as the new standard

Customer service is undergoing a revolution thanks to AI. The blog post on the topic ‘From call centre to AI headquarters’ describes this transformation. The AI Act has a noticeable impact here: customers must be informed when they interact with AI systems. This means a small amount of additional effort during implementation, but also the opportunity to build trust through transparency.

Virtual power plants: innovation with responsibility

Without AI, it would be almost impossible to bundle decentralised energy sources into virtual power plants. Many applications, such as load and generation forecasts, trade optimisation and market analysis, fall into lower risk categories under the AI Act. Here, transparency requirements are particularly important, so that companies can use these systems with relative ease.

The situation is different when AI is directly involved in the control and stabilisation of critical network interventions. In this area, the classification is considered high-risk AI, with extensive requirements for documentation, risk management, testing and monitoring procedures, and conformity assessment. These requirements will be binding from August 2027. Those who overcome these hurdles will not only achieve regulatory certainty, but also gain a clear competitive advantage through robust, trustworthy systems.

Load forecasting and grid monitoring: Unbridled market opportunities

Accurate load forecasts support the use of volatile renewable energies. Such forecasting applications generally fall into low-risk categories, provided they do not automatically intervene in safety-critical control systems. Nevertheless, transparency and GPAI obligations must be observed. AI-based approaches that intelligently link meteorological data, consumption patterns and market data also usually fall into lower risk categories. Energy companies can get started here without major regulatory hurdles.

AI-supported grid control, on the other hand, is caught between enormous potential and high regulatory requirements. Due to its relevance to critical infrastructure, it can be classified as high risk and is subject to comprehensive documentation, testing and monitoring obligations.

An AI-based inspection of smart meter installations is a typically less critical use case that benefits from lower regulatory hurdles.

Predictive maintenance: improving quality through regulation

AI-supported maintenance prediction shows how the AI Act is becoming a driver of quality. Our blog post on the topic of ‘Using AI for efficient and fail-safe power plant maintenance’ illustrates the potential. Systems used in critical infrastructure are considered high-risk and are therefore subject to stricter requirements. From documentation and testing procedures to transparency and monitoring requirements, there are a number of conditions to be observed. These requirements lead to more robust, trustworthy solutions and increase the quality of the AI systems used, even in non-high-risk areas.

Analysis and consulting solutions: the silent winners

Thanks to favourable conditions for business intelligence applications, market analysis tools and strategic consulting solutions, these benefit indirectly from the AI Act. While demand for data-driven decision-making aids is continuously increasing, these technologies benefit from a favourable regulatory status: most BI and analysis tools fall into the ‘low risk’ category and are therefore subject to minimal compliance requirements. Nevertheless, providers must comply with basic transparency requirements, in particular the clear labelling of AI-generated content. Stricter requirements apply if the systems are based on GPAI or are used in particularly sensitive business areas. This regulatory relief enables companies to focus more resources on innovation and user-friendliness instead of costly certification processes, and to bring their solutions to market faster.

The reality of the AI Act: brake or accelerator?

Where challenges arise

High-risk systems in critical infrastructures face significant compliance hurdles. From August 2027, comprehensive risk assessments, quality management systems and continuous monitoring will be mandatory. This means that, as things stand today, companies must prepare for longer development cycles and higher project costs for compliance measures.

Where opportunities await

Low-risk applications benefit from legal certainty. The required transparency leads to more robust, traceable systems. Companies that successfully overcome compliance hurdles can position themselves as trustworthy partners.

Setting the course now

At our last GenAI Utilities User Group meeting on 22 August, industry experts discussed practical experiences and solutions for all AI topics in the energy industry under the motto ‘DX goes GenAI’.

You can find further inspiration for the use of AI in the energy industry in our overview of GenAI in the energy industry.

Disclaimer:

The information provided in this article has been compiled with the utmost care and is for general information purposes only. It does not constitute legal advice or a binding interpretation of the AI Act or other legal norms.
Legal frameworks are subject to change, and the interpretation of laws always depends on the individual case. For binding legal assessments, please consult a qualified solicitor.
Neither the author nor adesso SE accept any liability for the accuracy, completeness or timeliness of the content provided. Any use of the information is at your own risk.