Gains:
- Ability to recognize types of greenwashing (vague, unsubstantiated, secret exchange, irrelevant, exaggerated future, false label) and make a claim measurable, sourced and scoped.
- Ability to test the defensibility of claims such as 'carbon neutral' by putting them through the 'reduce first, then offset with verified credit' test
- Ability to use artificial intelligence as a tool that scans the text with the eye of a claim checker, rather than producing greenwashing
The most dangerous trap in sustainability communication comes from good intentions rather than bad intentions: a company is genuinely environmentally conscious but describes its impact in exaggerated, vague, or unsubstantiated language. This is called greenwashing: making a product, service or company appear more environmentally friendly than it is. It used to be just a reputational risk; Today, it is a crime that gives rise to fines and legal liability under EU and national consumer legislation. In this unit, you will learn to use artificial intelligence (AI) as a tool to catch greenwashing, not produce it.
The critical irony is that AI is inherently very prone to producing greenwashing language. Since good-sounding phrases like "environmentally friendly", "environmentally friendly", "green", "carbon neutral" appear abundantly in the training data, the AI eagerly suggests them. That's why AI can be both your greatest source of greenwashing and (when used correctly) your best greenwashing detector. The difference is in what you ask him.
Common types of greenwashing
Greenwashing doesn't come in just one form. Recognizing the major types is the first step to catching it:
Genre
What do you mean?
example
uncertainty
Immeasurable, vague expression
"Environmentally friendly product"
lack of evidence
No number/source
"We are carbon neutral" (uncertified)
secret exchange
Highlighting the good and hiding the bad
"Recyclable" (but very dirty to produce)
irrelevance
True but trivial claim
"CFC-free" (already banned)
extravagant future
distant, unbinding promise
"Net zero by 2050" (no intermediate target)
fake label
Made-up/self-issued certificate
"Green approved" (who approved it?)
A good environmental claim has three characteristics that are the antidote to greenwashing: measurable (contains a number), proven (based on evidence), and scoped (it is clear what is being compared, to what, and when).
Tip: To test a claim, ask three questions: (1) Can I express this with a number? (2) Where is the evidence? (3) What am I comparing with and according to which date? If there is no response to all three, that sentence risks greenwashing and should be corrected before it goes into the report.
The trap of "carbon neutral" and similar claims
Two statements in particular require special attention. Carbon neutral claims that a company measures its emissions and compensates for the remainder through carbon offsetting (buying emissions-reducing/removing credits elsewhere). The problem is: if it's called "neutral" with unverified credits, without mitigation first, that's greenwashing. The EU introduces regulations that limit unsubstantiated "carbon neutral" claims. The rule: reduce first, then offset the inevitable remainder with verified credits and explain this transparently.
Weak prompt / Strong prompt
Weak prompt:
Write environmentally friendly and impressive marketing sentences for our product.
AI produces statements such as “completely eco-friendly, natural, green” that are immeasurable and carry legal risk.
Powerful prompt:
Your role: greenwashing auditor and honest communication expert. Task: write greenwashing-free, measurable and sourced claim statements with VERIFIED data below. Data: 60% of packaging recycled (2024, internal measurement); production emissions reduced by 18% from 2023 (third party verification available). Rules:- Eliminate vague/unsubstantiated statements such as “eco-friendly, natural, green, carbon neutral” USING.- Add number, source and comparison period to each claim.- Narrow the scope: write down what improved, relative to what, and when.
Using AI as a greenwashing detector
The real value of AI is to scan your text with the eye of an “environmental claim checker”. This is the most important check before publication.
Check the following marketing/report copy for greenwashing.For each environmental claim:1) Is it measurable? (is there a number)2) Is it welded? (is there evidence)3) Is its scope clear? (according to what/when)4) Which type of greenwashing does it fall into? (vague/no evidence/secret exchange/irrelevant/hyped future/fake label)Mark each risky statement and suggest a concrete, measurable alternative.Text: [TEXT]
The following claim tests the defensibility of a “carbon neutral” claim.
Your role: climate claims auditor.Question the following “carbon neutral / net zero” claim:- Were emissions measured and reduced first, or were offsets used directly?- Are offset credits validated/permanent?- Is scope (1/2/3) specified?- Is there an interim target, or is it just a distant date?Mark any missing points as “INDEFENDABLE — must be corrected”.Claim: [CLAIM]
Beware: Greenwashing happens not only in marketing slogans, but also in the official sustainability report. Sentences such as "We are the most sustainable company in the industry" and "minimum impact on nature" also carry legal risks in a CSRD report. Audit and legislation cover every sentence of the report.
three mini cases
Case 1 — From vague to concrete. One cosmetics brand said it was "completely natural and environmentally friendly." The AI audit marked this as “unclear + no evidence.” The wording was changed to “92% of the ingredients are of plant origin (2024, supplier certified)”; It was both true and defensible.
Case 2 — Carbon neutrality without evidence. One company declared, “We will be carbon neutral in 2024”; However, he had made no reductions and had simply purchased cheap and unverified credits. AI claim audit flagged this as "INDEFENSIBLE". The company withdrew the claim and announced a transparent roadmap, first with a mitigation plan and then with verified credits.
Case 3 — Secret exchange. One packaging company claimed it was “100% recyclable,” but the production of the product was very high-emitting and practically no facilities were recycling it. The audit caught the claim as a “secret exchange”; The statement was corrected to "recyclable (in regions with suitable facilities)" and production emissions were reported separately.
Common mistakes
- Getting the AI to print "impressive" text. This demand directly generates greenwashing; ask for “measurable and sourced” instead.
- Claiming “carbon neutral” without reducing it. Decrease first, then offset with verified credit; The opposite cannot be defended.
- Relying on vague adjectives. “Eco-friendly, green, natural” is immeasurable and legally risky.
- Ignoring the secret exchange. It is misleading to highlight one good thing and hide a great bad.
- Thinking the report is exempt from marketing. Greenwashing rules also include sentences within the official report.
In summary
Greenwashing often arises not from malice but from careless and exaggerated language; but today the result is fines and legal liability. The AI is very prone to producing this language, so never make it write "impressive environmental text". Instead, use AI as a greenwashing detector: scan each claim for measurability, source, and scope, identify its type, and replace it with a concrete alternative. A good environmental claim is always measurable, sourced and scoped. Publish sentences that are "provable", not "sounds good".
Application task
Write a short marketing copy that includes at least three questionable phrases (“eco-friendly,” “carbon neutral,” “100% recyclable”). Ask the AI to flag each claim with its type with the greenwashing audit prompt above. Then: (1) replace each risky statement with a measurable and sourced alternative, (2) test the “carbon neutral” claim, if any, for defensibility, (3) make honest a statement that involves an implicit trade-off.
checklist
- [ ] I tested each environmental claim for measurability, source, and scope.
- [ ] I weeded out vague phrases like “eco-friendly, green, natural.”
- [ ] I have put "carbon neutral" claims through abatement-first and verified credit testing.
- [ ] I corrected the expressions containing hidden trade-offs (highlighting the good and hiding the bad).
- [ ] I replaced or removed the fake/self-issued tags with the real certificate.
- [ ] I checked the claims with the same care in both the marketing and the official report.
- [ ] I used AI as a claim checker, not a text generator.