Wonder & Worry

Guidelines by field

AI guidelines on environment and energy what the field's own bodies ask

The environmental field is still at the measurement stage. There is broad agreement that AI's footprint should be counted, across training, use, hardware and buildings, but not yet on how. Standards bodies and the EU are building the methods now.

Last reviewed 27 October 2026. Every claim links to its source.

At a glance

  1. Measure the whole life cycle

    Training, use, hardware and data centers, not just one query.

  2. Report comparably

    With a stated method.

  3. Ask whether AI is needed at all

    Compare it with a lower-resource option first.

  4. Watch the rebound

    More efficient AI can lead to more total use and more total impact.

  5. Make only verifiable claims

    And remember water as well as electricity.

Who has spoken

Known or estimated energy consumption of the model.

EU AI Act, Annex XI

Binding and voluntary

Binding

EU documentation duties for general-purpose model providers and EU data-center reporting. No US federal requirement to disclose AI energy or water use was found.

Voluntary

ISO/IEC TR 20226, SCI for AI, AFNOR's Frugal AI and companies' own disclosures.

What changed in 2025 and 2026

Questions to ask yourself

  • Does this task need AI, or would a search, a template or my own draft do?
  • Am I generating many versions I will throw away?
  • When a company tells me its AI is "green", what method and scope sit behind the number?

How this lines up with my everyday guidelines

My everyday guidelines beside this field's nearest rules
My everyday guidelinesThe field's nearest rule

The Shared Ethic: Use energy with intention.

Ask first whether AI is needed, and measure the whole life cycle (AFNOR; ISO). Energy is not part of my everyday guidelines; the Shared Ethic covers it.

6. Stay curious and critical Question outputs. AI can be confident and wrong.

Make only verifiable environmental claims (AFNOR).

4. Use for good Make sure uses align with your values and professional ethics.

Sustainability and environmental health as a core value (UNESCO).

Source 289

International Organization for Standardization. ISO/IEC TR 20226:2025, Environmental sustainability aspects of AI systems. Jul 2025. iso.org/standard/86177.html

See all sources

Source 290

Green Software Foundation. SCI for AI specification ratified: a standard for measuring AI emissions. Dec 2025. greensoftware.foundation/articles/sci-ai-specification-ratified-standard-for-measuring-ai-emissions-across-the/

See all sources

Source 291

AFNOR. Reference framework to reduce the environmental impact of AI (Spec 2314, Frugal AI). afnor.org/en/news/artificial-intelligence/reference-framework-reduce-environmental-impact-ai/

See all sources

Source 287

EU AI Act, Annex XI (technical documentation for general-purpose AI models). artificialintelligenceact.eu/annex/11/ ; sota.io (third-party summary). EU AI Act energy consumption reporting consultation. 2026. sota.io/blog/eu-ai-act-gpai-art53-energy-consumption-reporting-consultation-developer-guide

See all sources

Source 288

DLA Piper. New data centre sustainability reporting obligations introduced through EU Regulation 2024/1364. Sep 2024. dlapiper.com/en/insights/publications/2024/09/new-data-centre-sustainability-reporting-obligations-introduced-through-eu-20241364-regulation

See all sources

Source 293

CBS News. AI environment impact study: energy usage of a Google Gemini prompt. Aug 2025. cbsnews.com/news/ai-environment-impact-study-energy-usage-google-gemini-prompt

See all sources

Source 85

Commission Delegated Regulation (EU) 2024/1364 of 14 Mar 2024, on a common Union rating scheme for data centres. eur-lex.europa.eu/eli/reg_del/2024/1364/oj/eng

See all sources

Source 292

G7 French Presidency. Overview of initiatives for measurement and monitoring of energy and resource use of AI models. 29 May 2026. entreprises.gouv.fr/files/files/Actualites/2026/g7/overview-measurement-monitoring-energy-resource-ai-models.pdf

See all sources