Chapter 09 · Sustainability & AISustainability & AI
EU AI Act
Definition
Regulation (EU) 2024/1689 — the European Union's comprehensive, risk-based law on artificial intelligence, in force since 1 August 2024 with obligations phasing in to 2028.
References
This source provides part of the technical or institutional basis for the definition.
This source supports the explanation of how the term is applied, measured or governed in practice.
Overview
What it means
The Act bans a small set of unacceptable-risk practices, imposes requirements on high-risk systems, and creates transparency duties for general-purpose AI models — including documenting known or estimated energy consumption (Annex XI).
Environmental provisions are otherwise modest: Article 40 tasks standards work with AI resource performance, and Article 95 encourages voluntary codes covering environmental sustainability. A 2026 'Digital Omnibus' deferred some high-risk deadlines.
How it is used
Providers and deployers map their systems to the Act's risk tiers; GPAI model providers compile Annex XI technical documentation; the European AI Office supervises GPAI obligations. Penalties reach €35 million or 7% of turnover for banned practices and €15 million or 3% for most other violations.
Why it matters
The AI Act is the first comprehensive horizontal AI law, and its treatment of energy documentation — however limited — is the first time AI's environmental cost has entered binding tech regulation. Its gaps (no carbon accounting, authority-only disclosure) define the frontier for follow-up policy.
Current status
GPAI obligations apply since 2 Aug 2025; Digital Omnibus (final Council approval June 2026) defers high-risk duties to Dec 2027/Aug 2028.