In short
Claim only what you have measured, be specific about what improved and against what baseline, and state the method, boundary and limits. Separate efficiency per task from total use, and don't rest a claim on offsets or annual certificates. Vague words such as "green AI" or "carbon-neutral AI" are risky, and in the UK and EU can breach consumer law.
Steps
- Write the claim as a specific, measured statement with a baseline and a date.
- Keep the method, boundary and data behind the claim ready to share.
- Say whether the improvement is per task or in total, and show both if you can.
- Remove generic words such as "green" or "eco-friendly" unless the evidence supports them.
- Check the wording against the consumer rules where you sell before publishing.
A safe claim is specific, measured and comparable; a risky one is vague. "Our document summariser now uses about half the energy per document it used in January, measured on the same hardware" can be checked. "Sustainable AI" cannot.
What do the rules say?
- UK: the Competition and Markets Authority's Green Claims Code sets six principles, including that claims must be truthful, clear and substantiated. Since April 2025 the CMA can fine up to 10% of global turnover for consumer law breaches.
- EU: Directive (EU) 2024/825 has applied since 27 September 2026. It bans generic environmental claims without proof and product claims of climate neutrality based on offsetting.
- US: the Federal Trade Commission's Green Guides set out how environmental marketing claims are judged.
What should a claim include?
- What improved: energy, emissions or water, per what unit.
- Against what: the baseline and when it was measured.
- How: measured or estimated, tool and boundary.
- The limits: what is left out, such as training or users' devices.
The Green Software Foundation's Software Carbon Intensity specification is a useful model: it does not let offsets or renewable certificates reduce the score.
Common mistakes
- Presenting a per-request efficiency gain as a cut in total emissions while use grows.
- Calling a service "carbon neutral" because of offsets.
- Quoting a supplier's figure without saying it is theirs.
See honest green claims for worked examples.
Sources
- Green Claims Code: making environmental claims, Competition and Markets Authority (2021)
- Directive (EU) 2024/825 on empowering consumers for the green transition, EUR-Lex (2024)
- Guides for the Use of Environmental Marketing Claims (16 CFR Part 260), US Federal Trade Commission (eCFR) (2012)
- Software Carbon Intensity specification, Green Software Foundation (2024)
Related
Honest green claims about AI
How to avoid greenwashing about AI: what you can and can't claim, rebound effects, honest reporting, and the UK, EU and US rules on environmental claims.
How do I spot greenwashing in AI marketing?
Warning signs of greenwashing in AI products and services, from vague 'sustainable AI' labels to offset-based neutrality claims, and the questions to ask.
How do I set a baseline to prove AI energy savings?
Show a real AI energy saving: measure before any change on the same set-up and load, repeat runs, report per unit of work, and check quality held.
How do I report AI emissions in our sustainability report?
Where AI belongs in Scope 2 and Scope 3 reporting, how to estimate it when suppliers give little data, and how to state method and uncertainty.