# 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.

Source: https://greenai.guide/how-do-i/spot-greenwashing-in-ai-marketing/  
Updated: 2026-11-19

## In short

Be wary of words like "green", "sustainable" or "carbon-neutral AI" without figures, methods and boundaries behind them, and of claims that rest on offsets or annual renewable certificates. Honest claims say what was measured, against what baseline, and what was left out; in the EU, generic environmental claims that cannot be proved are banned.

## Steps

1. Ask what exactly was measured and what the figure includes.
2. Check whether a claimed saving is per task or in total.
3. Look for the baseline the claim compares against.
4. Treat "carbon neutral" claims based on offsets or certificates with caution.
5. Check whether the figure is self-reported or independently verified.

**Greenwashing is a green claim that can't be checked.** As AI's energy use attracts attention, more products are
marketed as "efficient", "sustainable" or "green AI". Some are; the test is whether the claim comes with evidence.

## What are the warning signs?

- **Vague labels.** "Eco-friendly AI" or "sustainable intelligence" with no numbers.
- **Neutrality through offsets.** "Carbon neutral" because credits were bought, not because emissions fell. The
  Green Software Foundation's Software Carbon Intensity specification does not let offsets or renewable
  certificates reduce a software carbon score.
- **Efficiency presented as reduction.** "50% less energy per query" says nothing about total use, which may still
  be rising. Luccioni, Strubell and Crawford argue that efficiency gains in AI can be offset by growing use, the
  [rebound effect](/glossary/#rebound-effect).
- **No boundary.** A per-prompt figure that counts only the chips looks better than one that includes cooling and
  idle machines.
- **Cherry-picked comparisons** against an old or unusually inefficient baseline.

## What does the law say?

**In the EU, the rules have tightened.** Directive (EU) 2024/825 bans generic environmental claims to consumers that
cannot be proved. Other countries have their own consumer-protection rules on misleading claims; check your
national regulator's guidance.

## What questions should I ask?

1. What was measured, and what was left out?
2. Per what: per request, per user, per year?
3. Compared with what baseline?
4. Who measured it, and can it be checked?

## Common mistakes

- **Assuming a renewable-energy pledge** makes every use of a service green.
- **Treating "AI for good" as "green AI".** Using AI for a climate goal does not make the AI itself low-impact.

If you're the one making claims, see [make honest claims about green AI](/how-do-i/make-honest-claims-about-green-ai/)
and the full guide to [honest green claims](/practice/honest-green-claims/).

## Sources

- [Directive (EU) 2024/825 on empowering consumers for the green transition](https://eur-lex.europa.eu/eli/dir/2024/825/oj), European Union (2024)
- [Software Carbon Intensity specification](https://sci.greensoftware.foundation/), Green Software Foundation (2024)
- [From efficiency gains to rebound effects: the problem of Jevons' paradox in AI's polarized environmental debate](https://arxiv.org/abs/2501.16548), Luccioni, Strubell and Crawford, ACM FAccT (2025)
