Pinned snapshot. This article is rendered from the catalog state captured at (closest snapshot at-or-before your requested date of ). The live version may differ — drop the
?asOf= parameter to see the current catalog state.Energy consumption, water usage, carbon emissions, and resource demands of large-model training + inference. EU AIA Recital 142 + Art. 95 voluntary codes; NIST AI 600-1 Environmental Impacts (named risk category); G7 Hiroshima Code §6 sustainable AI; emerging French ARCEP + Spanish AI Bill obligations; SDG-linked references in UN + AU + ASEAN frameworks.
Definition & scope
The cross-jurisdiction picture below shows how each of 45 tracked instruments treats this topic. The patterns vary substantially — and 33 regimes are silent, leaving gaps that future policy work could address.
Coverage across jurisdictions
Historical primacy & cross-jurisdiction tension
First addressed by OECD AI Principles (Recommendation) on (implicit). Subsequent regimes have either codified, diverged from, or remained silent on this baseline.
Compare jurisdictions: EU vs US · EU vs UK · EU vs CN
Enforcement & impact
Silent regimes — gap signal
Instruments that do not address Environmental Impact of AI Training — candidates for future policy work.
- Executive Order 14179 — Removing Barriers to American Leadership in AIUS
- UK Pro-Innovation Approach to AI Regulation (White Paper)UK
- Interim Measures for Generative AI Service ManagementCN
- NIST AI Risk Management FrameworkUS
- Bletchley Declaration on AI Safetyglobal
- Seoul Declaration on Safe, Innovative and Inclusive AIglobal
- California SB-1047: Safe and Secure Innovation for Frontier AI Models ActUS
- India Digital Personal Data Protection Act + AI Advisory (MEITY)IN
- Brazil AI Bill (PL 2338/2023)BR
- Anthropic Responsible Scaling Policy (RSP) v2US
- OpenAI Preparedness FrameworkUS
- Google DeepMind Frontier Safety FrameworkUS
- Meta Frontier AI FrameworkUS
- UK-US AI Safety Institute Memorandum of Understandingglobal
- White House Voluntary AI CommitmentsUS
- Singapore Model AI Governance Framework for Generative AISG
- Japan METI AI Guidelines for BusinessJP
- General Data Protection Regulation (GDPR)EU
- EU General-Purpose AI Code of PracticeEU
- OMB Memorandum M-24-10 (Advancing Governance, Innovation, and Risk Management for Agency Use of AI)US
- GSA Generative AI and Specialized Computing Infrastructure Acquisition Resource GuideUS
- DoD Responsible AI Strategy and Implementation PathwayUS
- FedRAMP AI Cloud Procurement GuidanceUS
- DFARS Subpart 252.204 (Safeguarding Covered Defense Information and Cyber Incident Reporting)US
- California SB-53: Transparency in Frontier Artificial Intelligence Act (TFAIA)US
- California SB 243: Companion ChatbotsUS
- California SB 942: AI Transparency ActUS
- Revised Product Liability Directive (Directive (EU) 2024/2853)EU
- Directive (EU) 2024/2831 on improving working conditions in platform workEU
- Provisions on the Administration of Deep Synthesis of Internet Information ServicesCN
- New York RAISE Act: Responsible AI Safety and Education ActUS
- TAKE IT DOWN Act (Tools to Address Known Exploitation by Immobilizing Technological Deepfakes on Websites and Networks Act)US
- Japan AI Promotion Act (Act on the Promotion of Research, Development and Utilization of AI-Related Technologies)JP
See also
Further reading
13 academic & grey-literature sources bearing on this topic — catalogued metadata with a primary link; one-line findings are ✦ AI-generated summaries, labeled as such (charter §7.9). Browse the full literature index.
- AI, Climate, and Regulation: From Data Centers to the AI Act Peer-reviewed✦ AIAnalyses the legal levers (AI Act energy-reporting duties, Energy Efficiency Directive data-centre KPIs, sustainability reporting) for governing AI's climate footprint and their disclosure gaps.
- Making AI Less 'Thirsty': Uncovering and Addressing the Secret Water Footprint of AI Models Peer-reviewed✦ AIEstimates training GPT-3 in US data centres can evaporate ~5.4 million litres of water and projects 4.2-6.6 billion m3 of AI water withdrawal by 2027, arguing water use needs reporting and scheduling.
- Power Hungry Processing: Watts Driving the Cost of AI Deployment? Peer-reviewed✦ AIMeasures deployment energy/carbon per 1,000 inferences, finding 'multi-purpose, generative architectures are orders of magnitude more expensive than task-specific systems.'
- The growing energy footprint of artificial intelligence Peer-reviewed✦ AICanonical estimate projecting AI servers could consume 85-134 TWh/year by 2027 (comparable to a small country), framing disclosure of AI electricity use as a policy problem.
- Estimating the Carbon Footprint of BLOOM, a 176B Parameter Language Model Peer-reviewed✦ AILife-cycle estimate finding BLOOM's training emitted ~24.7 tCO2e from dynamic power but ~50.5 tCO2e once manufacturing and idle/operational consumption are counted, motivating full-lifecycle reporting.
- The Carbon Footprint of Machine Learning Training Will Plateau, Then Shrink Peer-reviewed✦ AI'Four best practices can reduce ML training energy by up to 100x and CO2 emissions up to 1000x'; predicts training's total carbon footprint will plateau, then shrink.
- Aligning artificial intelligence with climate change mitigation Peer-reviewed✦ AIPresents 'a systematic framework for describing the effects of machine learning (ML) on GHG emissions' and suggests 'policy levers' for shaping ML's climate impacts.
- Chasing Carbon: The Elusive Environmental Footprint of Computing Peer-reviewed✦ AIShows embodied (manufacturing) carbon can rival operational emissions for computing systems, grounding the case that AI footprint accounting and rules must include hardware lifecycle, not just training energy.
- Green Algorithms: Quantifying the Carbon Footprint of Computation Peer-reviewed✦ AIProvides a standardized, reproducible methodological framework (and calculator) to estimate the carbon footprint of any computational task from runtime, hardware and grid location.
- Carbon Emissions and Large Neural Network Training Preprint✦ AIComputes energy and carbon for T5, Meena, GShard, Switch Transformer and GPT-3, showing operational choices (model, datacentre, hardware, region) can shift training emissions by orders of magnitude.
- Green AI Peer-reviewed✦ AICoins 'Green AI', arguing compute/energy efficiency should be reported as a first-class evaluation metric alongside accuracy to curb the rising environmental cost of deep learning.
- The carbon impact of artificial intelligence Peer-reviewed✦ AISurveys evidence that ML's carbon cost is under-measured and calls for tools to quantify training footprints and a shift to sustainable AI infrastructure as a governance priority.
- Energy and Policy Considerations for Deep Learning in NLP Peer-reviewed✦ AICanonical policy paper 'quantifying the approximate financial and environmental costs of training' NLP models, with 'actionable recommendations to reduce costs and improve equity.'
References
The primary instrument sources behind the article's classifications.
- EU-AIA-2024: Art. 95 voluntary codes of conduct include environmental sustainability; Recital 142 references energy efficiency reporting for GPAI
- US-EO-14110: §5.2 directs environmental-review consideration; §4.2 reporting includes some energy data
- G7-HIROSHIMA: Code §6 references sustainable AI development; not detailed obligation
- OECD-AI-PRIN: Principle 1.1 inclusive growth + sustainable development; addresses environment implicitly
- COE-AI-CONV: Art. 7 sustainability principle; environmental impact subsumed
- UN-RES-2024: Preamble references SDGs which include climate goals
- NIST-AI-RMF-GENAI: NIST AI 600-1 — Environmental Impacts is one of 12 named GenAI risk categories
- ASEAN-AI-GUIDE-2024: Guide references sustainable AI principles; not operationalised
- AU-AI-STRATEGY-2024: Continental strategy includes sustainability themes; not operationalised
- UNESCO-AI-ETHICS-2021: Policy Area 'Environment and Ecosystems', para 84 — assess direct/indirect environmental impact incl. carbon footprint + energy consumption
- IT-AILAW-2025: Art. 3(1) lists 'sostenibilità' (sustainability) among the binding general principles governing AI development and use, alongside transparency, proportionality, security and non-discrimination. No operative environmental-reporting or training-footprint duty.
- UN-GDC-2024: GDC para 11(e) lifecycle sustainability; Objective 5 narrative (A/RES/79/1, Annex I)
Cite this article 8 formats · BibTeX, RIS, APA, Chicago, … · 1-click copy
Verify the year + paste-and-refine. Primary source linked in BibTeX/RIS note.
Persistent identifier: https://policywindow.org/wiki/environmental-impact-of-training — committed-stable URL with content-versioning via ?asOf= (rollout pending per methodology §7). DOIs via Zenodo are on the roadmap.
Article tools — track changes, suggest an edit
View history — every captured revision of this article · What links here
Source: Edit on GitHub (search for `environmental_impact_of_training`)
12 instruments tracked.