Suggest compliant wording: the green claim rewriter
Last updated June 12, 2026
What it does
The editor continuously scans your passport's free-text fields for environmental-claim language that is risky under the EU's green claims rules (the ECGT Directive — background in our guide to the EU green claims ban). When it finds some, it raises a non-blocking advisory, and the rewriter offers a way to act on it: AI-generated rewrites of your own text that keep your message while removing or qualifying the risky claim.
The scanning itself is deterministic, not AI — a fixed, locale-aware term list that behaves identically every time. AI is only involved when you ask for rewrite suggestions.
When the button appears
Suggest compliant wording shows under a field when all three are true:
- The field is one of the prose fields a rewrite makes sense for: Product Name, Product Description, the free-text care instructions, Recycling Instructions, or Disposal Instructions.
- The field's current text carries a claim advisory — shown as a quiet "N suggestions for your environmental claims" line under the field.
- The field isn't empty.
What the claim scanner flags
The scanner recognizes these categories of risky language, in all eight interface languages:
| What it spots | Examples | Why it's flagged |
|---|---|---|
| Generic environmental claims | "eco-friendly", "biodegradable", "green product", "conscious collection", "sustainable materials" | Generic claims need their basis stated in clear and prominent terms on the same medium, or backing by a recognized scheme such as the EU Ecolabel. Note: bare words like "green" or "conscious" are only flagged when combined with product-level qualifiers (e.g., "green product", "conscious collection") |
| Carbon-neutrality claims | "climate neutral", "carbon neutral", "net zero", "climate positive" | Offset-based neutrality claims are banned outright; own-lifecycle claims must be specified and independently substantiated |
| Comparative claims | "greener than", "more sustainable than" | Comparisons must disclose their method and assumptions |
| Legal requirements as features | "PFAS-free", "azo-free", listing REACH as a distinction | Presenting what the law requires of every product in the category as a distinctive feature is prohibited |
| Forward-looking claims | "net zero by 2030" | Future claims need a public, objective implementation plan with independent verification |
| Whole-product overstatement | "made from recycled materials" with recycled content below 100% | A whole-product claim must not be made when it only applies to part of the product |
These all surface as advisories — they never disable the Publish button. The difference between errors and advisories is explained in Completeness and validation.
Review and apply
- Click Suggest compliant wording under the flagged field. The dialog "Compliant wording suggestions" opens and generates fresh candidates from your current text.
- Read the explanation of why your original wording is risky, then the candidate rewrites.
- Each candidate is re-checked by the same deterministic scanner and labelled either "Clears the advisory" or "Lower risk — still review". A candidate is never offered if it would introduce a new risky claim.
- Click Use this wording on the candidate you want (it replaces the field's text and saves like any other edit), or close the dialog to keep your original.
The dialog states its own terms, and they're worth taking literally: "AI-generated from your own text — review and edit before saving. You remain responsible for your claims." Applying a suggestion makes it your statement; the same responsibility attaches to it as to anything you typed yourself.
Usage allowance
AI suggestions are metered per organization, not per user. All AI assists share a single allowance of 100 generations in any rolling 24-hour window — the claim rewriter and the care symbol suggester draw from the same pool. Each generation counts the moment it runs, whether or not you apply the result: opening the dialog generates one suggestion set, and Try again generates another. When the allowance is used up, requests fail with "Couldn't generate suggestions. Please try again.", and start succeeding again as earlier generations fall outside the 24-hour window.
Generating suggestions requires the editor role or higher; viewers don't get the button's effect.
What it is not
Not legal advice, and not a compliance guarantee
The rewriter helps you avoid known-risky phrasings — it cannot tell you whether a claim is true or substantiated, and clearing an advisory is not a determination that your text is lawful. As the publish flow itself puts it: "This guidance is provided for information only and is not legal advice." For your specific obligations, talk to a qualified adviser.
Substantiation stays with you. If your passport contains environmental claims, publishing asks you to confirm you're responsible for their accuracy and lawfulness and hold the substantiation for them — rewritten or not.
Frequently asked questions
Does the rewriter change my text automatically?
No. It only proposes candidate rewrites in a dialog. Nothing touches your passport until you click 'Use this wording' on a specific candidate — and you can close the dialog and keep your original text.
Do claim advisories block publishing?
No. Claim-language advisories are warnings for your review, never blockers. You can publish with advisories open — though if your passport contains environmental claims, publishing asks you to confirm responsibility for them first.
Why do I sometimes get no suggestions?
When no compliant rewrite can be generated, the dialog says so: 'No compliant rewrite could be generated automatically. Try stating the basis of your claim, or remove it.' Some claims can't be fixed by rewording — they need substantiation or removal.