Analysis
The AI Act in construction and industry — what applies today and what has been postponed
The AI Business Center Foundation (AIBC) is a think tank and advisory partner that helps the boards of construction, industrial and technical-services companies adopt artificial intelligence so that it delivers measurable results and builds competitive advantage. This analysis sets out what the AI Act means for construction and industrial companies after the changes of 2026.
Fear of penalties is no longer a good adviser. The timetable has become one
In spring 2025 we wrote that the AI Act would change the way construction and industrial companies operate. The thesis still holds, but its details needed correcting. Regulation (EU) 2024/1689 has been amended by Regulation (EU) 2026/1744, known as the AI Omnibus, which entered into force on 27 July 2026. The timetable has changed, and in 2025 we ourselves simplified several matters more than was justified.
What applies today
- Since 2 February 2025: prohibited practices (Art. 5) and AI literacy (Art. 4). After the AI Omnibus, providers and deployers of AI systems must support the development of AI literacy of their staff rather than ensure a set level. The provision does not require the purchase of training.
- Since 2 August 2025: the rules on general-purpose AI models, governance and penalties.
- Since 2 August 2026: the AI Act applies in general, including the transparency obligations of Art. 50. For a construction or industrial company this most often means informing customers at first contact that they are dealing with an AI assistant, and clearly labelling images, recordings and video created using AI where they could mislead the audience as to whether the persons, objects, places or events shown are real.
What has a date
| date | what starts to apply |
|---|---|
| 2 December 2026 | end of the transition period for machine-readable marking of content by providers of AI tools |
| 2 August 2027 | obligations of providers of general-purpose AI models placed on the market before 2 August 2025 |
| 2 December 2027 | obligations for high-risk systems under Annex III |
| 2 August 2028 | obligations for high-risk systems under Annex I, that is, those built into regulated products |
High-risk systems: correcting our 2025 thesis
In the first version of this text we listed fleet management, production quality control and inventory planning as examples of areas subject to the stricter regime. That was a simplification. Whether a system qualifies as high-risk depends on its intended purpose. Examples are systems used in recruitment, creditworthiness assessment, critical infrastructure and biometrics, as well as systems built into regulated products under Annex I. Some AI applications in the plant and on the construction site do not fall into this category, while others may, particularly systems that are components of regulated products under Annex I or that operate in critical infrastructure. We do not decide borderline cases in this analysis; a legal adviser assesses them on the basis of a description of the specific system.
For construction and industrial companies, the high-risk area of most practical relevance is usually recruitment, because AI tools for screening candidates reach HR departments faster than they reach production.
The role determines the obligations
The AI Act distinguishes between the provider of an AI system and the deployer. Most construction and industrial companies are deployers: they use systems supplied by others. Their scope of obligations then differs from a provider’s. A company that substantially modifies a system or places it on the market under its own name may, however, become a provider itself. That is why the first step is not documentation but a register: which AI systems the company has, what they are used for and who supplied them.
What to do in 2026
- List the AI uses in the company, including those that employees rely on without a formal decision.
- Establish the company’s role in each use.
- Review customer touchpoints with regard to Art. 50.
- Document the measures taken to support the AI literacy of staff.
- Plan measures for the obligations dated 2027 and 2028, where they concern the company.
Transparency is covered in detail on the page AI Act in communication and sales. Preparing the organisation is described on the page AI Act and ESG compliance.
What has changed since first publication
- We took account of the AI Omnibus (Regulation (EU) 2026/1744) and the new timetable: the obligations for high-risk systems have been moved to 2 December 2027 and 2 August 2028.
- We removed the sentence stating that “full responsibility” would fall on companies in 2026 and replaced it with a calendar of dates.
- We corrected the list of examples of high-risk systems and added the distinction between the roles of provider and deployer.
- We added Art. 4 as worded after the AI Omnibus and the transparency obligations of Art. 50, which apply from 2 August 2026.
- We removed the reference to a paid report and the “so as not to be caught out” tone. We bring clarity rather than alarm.
Frequently asked questions
Are predictive maintenance systems high-risk systems?
Not merely because they predict something. The high-risk category follows from a system's intended purpose, for example in recruitment, creditworthiness assessment, critical infrastructure or biometrics, or from being built into a regulated product under Annex I. A legal adviser assesses the specific case on the basis of a description of the system.
What in the AI Act already applies to a construction company?
Supporting the development of AI literacy of staff (Art. 4) since 2 February 2025, and the transparency obligations of Art. 50 since 2 August 2026, for example informing customers that they are dealing with an AI assistant. The obligations for high-risk systems have dates in 2027 and 2028.
Let’s start with a conversation
Let us talk about an AI use register for your company.