AI Transparency Disclosure

Last updated: April 14, 2026

EU AI Act 2024/1689 | Minimal Risk

In accordance with the EU Artificial Intelligence Act (AI Act, 2024/1689), DLV Digital S.A. discloses information about AI systems used in the CleanData service.

1. AI Systems We Use

CleanData uses the Claude large language model (Anthropic) for data normalization — correcting typos and standardizing formats for cities, names, and addresses. This is the only AI feature in the service.

2. How AI Works in CleanData

When you select AI normalization: (a) only unique values from selected columns are extracted from your file; (b) these values are sent to the Claude API (Anthropic, USA) for processing; (c) results are returned and applied to the file; (d) unique "original→normalized" pairs are cached to speed up future requests.

3. AI Is Opt-In Only

AI features are activated only when you explicitly select AI normalization. 80% of the service's features (duplicate removal, formatting, whitespace cleaning) work without AI — using only pandas algorithms. You can always use the service entirely without AI.

4. AI System Classification

Under the AI Act, our use of the Claude API is classified as a general-purpose AI system with minimal risk. The system does not make automated decisions affecting users' legal rights. AI normalization results are advisory: you review and confirm changes before downloading the file.

5. AI Provider

Provider: Anthropic, PBC (San Francisco, USA). Model: Claude. Anthropic processes data in accordance with its usage policy. We use API mode, under which Anthropic does not use submitted data for model training.

6. Limitations and Accuracy

AI normalization does not guarantee 100% accuracy. Possible errors include: incorrect normalization of rare names or addresses, ambiguous spellings. We provide a diff-view to review every change before applying. The user is responsible for the final data.

7. Your Data and AI

Only unique values from selected columns are sent to the AI system, not the entire file. Data is not retained by the provider (API mode). Our internal cache stores only anonymized value pairs without links to files or users.

8. Contact

Questions about AI usage: DLV_support@proton.me. You have the right to opt out of AI features and use the service with algorithmic cleaning methods only.