AI transparency
Wantwise uses artificial intelligence. This page explains how, in line with the EU Artificial Intelligence Act (Regulation (EU) 2024/1689), in particular its transparency obligations (Article 50).
What the AI does
- Understands your wish: turns what you write into 2–4 catalog search phrases and a one-line summary.
- Rates every product found: a verdict (worthy pick, questionable or not a fit), a score from 0 to 100, a short reason and a clean product name.
- It does not buy anything, contact anyone or decide for you: you choose what goes into your lists.
Which model
The texts are generated by Claude, a general-purpose AI model made by Anthropic, PBC. We are the deployer: we integrate the model into Wantwise and are responsible for how it is used here.
How AI content is marked
- On the page: AI ratings carry an “AI” label, and a note next to the results says which texts are written by AI.
- In the page code: AI-written elements carry the attribute data-ai-generated="true", and pages with AI content send the response header AI-Disclosure.
- Product photos, prices, shop names and the original product titles come from the merchants, not from the AI. The original title appears when you hover over the AI-written name.
Limitations
The AI judges only from the listing text, the price and the shop name. It does not see reviews, stock or the real product, and it can be wrong: it may miss a good product or rate a poor one too high. Prices and availability change. Always check the details in the shop before you buy.
Data
The model receives your wish, the market and the product listings, not your email address or name. Please do not type personal data about yourself or others into a wish. Under its commercial terms Anthropic does not use this data to train its models. More in the Privacy Policy.
Risk classification
The system recommends consumer products. It is not a high-risk AI system under Annex III of the AI Act, uses none of the prohibited practices of Article 5, performs no profiling, biometric identification or emotion recognition, and takes no decisions with legal or similarly significant effects on you.
Human oversight and feedback
Every AI rating is a suggestion: you see the original product, the shop and the price, and you choose. If an AI output is wrong, offensive or misleading, write to retuames@gmail.com; we review reports and adjust our instructions to the model.