The European Union's binding AI Act and the United States' voluntary, performance-based framework are pulling global construction projects in opposite regulatory directions, creating dual compliance burdens for contractors, equipment manufacturers, and modular builders operating across both markets.
Background
The EU AI Act (Regulation (EU) 2024/1689) is the first comprehensive legal framework for AI worldwide, establishing risk-based rules for AI developers and deployers regarding specific uses of AI.1EU AI Act News 2026: Compliance Requirements & Deadlines The Act entered into force on 1 August 2024, with prohibited AI practices and AI literacy obligations taking effect from 2 February 2025.2EU Regulation on AI | Insight | Baker McKenzie Rules for high-risk AI systems apply from 2 August 2026.
For construction specifically, regulatory exposure extends beyond the AI Act. The EU Machinery Regulation introduces rules for machinery incorporating AI and machine learning - if an AI system performs a safety function, the machine is classified as high-risk under Annex I. Two new categories have been added to the high-risk list for AI-enabled machinery: safety components with fully or partially self-evolving behaviour using machine learning, and machinery with embedded AI systems ensuring safety functions. Autonomous construction equipment, AI-driven collision avoidance systems, and real-time risk assessment platforms all fall within this scope.
In the United States, no equivalent binding federal rule exists for AI in construction. The NIST Artificial Intelligence Risk Management Framework (AI RMF), released by the U.S. National Institute of Standards and Technology, is a voluntary, explicitly non-binding tool for managing AI risks. OSHA's authority over AI-enabled construction tools remains anchored in its general duty clause and existing safety standards, with no AI-specific rulemaking currently in force.
Details
The EU compliance picture is defined by specific, mandatory obligations. Before placing a high-risk AI system on the EU market, providers must complete a conformity assessment demonstrating compliance with requirements spanning risk management, data quality, documentation and traceability, transparency, human oversight, accuracy, cybersecurity, and robustness. Third-party conformity assessments can cost between €10,000 and €40,000 per system, and documentation requirements for high-risk AI systems increase development time by 15% to 25%.
Enforcement timelines add further complexity. CEN and CENELEC have been unable to deliver required harmonised standards by the August 2025 deadline, and standardisation work remains ongoing. The Digital Omnibus proposal, published on 19 November 2025, linked the application of high-risk rules to the availability of support tools, setting the latest application date at 2 December 2027 for Annex III systems and 2 August 2028 for Annex I (product-embedded) AI systems. A political agreement on the AI Omnibus was reached on 7 May 2026.
In the United States, the approach diverges sharply. The U.S. government's AI strategy centres on reducing barriers to adoption and positioning the country as a global leader in AI infrastructure and innovation, with federal and state agencies encouraged to remove perceived regulatory bottlenecks. The U.S. AI Action Plan notably seeks to quash "burdensome" state regulation of AI. OSHA data illustrates the risk landscape on U.S. sites independent of AI governance: In 2025, OSHA recorded 5,914 violations for fall protection general requirements in construction alone, according to OSHA violation tracking data. In 2022, construction fatalities accounted for 21% of all workplace fatalities despite the sector employing only 7% of the workforce, according to the Bureau of Labor Statistics.
Some U.S. contractors are deploying AI safety tools under existing frameworks. Skanska introduced Safety Sidekick, an AI-powered assistant that delivers real-time, task-specific safety guidance to construction teams using Skanska and OSHA standards. In 2025, AI technology is shifting construction safety from reactive to proactive, with some companies reporting incident reductions of 40% to 50%. However, none of these deployments are subject to EU-style conformity assessment or mandatory incident reporting.
Penalties illustrate the asymmetry. Under the EU AI Act, infringements related to prohibited practices or data requirements carry penalties of up to €35 million or 7% of total worldwide annual turnover, whichever is higher; non-compliance with other requirements can reach €15 million or 3% of global turnover. The United States has no comparable AI-specific penalty structure for construction sites.
Outlook
ISO/IEC SC 42 is developing AI-related international standards, and EU alignment efforts are described as crucial to avoiding regulatory fragmentation and enabling AI developers to operate across multiple markets without redundant compliance. Multinational companies are adopting the NIST AI RMF as the "operational layer" beneath regulatory compliance, given its crosswalk alignment with EU AI Act quality management requirements. For contractors operating across both markets, ISO/IEC 42001 certification may simplify cross-jurisdictional operations, as ISO standards carry international recognition. Project managers and procurement leads sourcing AI-enabled safety analytics or autonomous plant equipment for EU-funded infrastructure programmes should begin conformity assessment planning now, given that high-risk AI system obligations under Annex III become enforceable on 2 August 2026, according to the European Commission.
This article references the existing post Construction Robotics, AI Face Data Standards Barrier for context on data standardisation challenges affecting AI adoption in construction.
