AI notice

Lovable Studio uses AI to help structure and surface the content you upload. It is decision support, not advice. The decision is always yours.

What AI does here

• Extracts text from your PDFs (with OCR fallback for scans).
• Proposes chapter titles, summaries, and a design profile.
• Answers reader questions grounded in your book content, with page citations.

What it doesn't do

• It does not make claims about facts outside your uploaded material.
• It does not tell readers what they "should" or "must" do.
• It does not learn from your data to train third-party models.

EU AI Act (Reg. 2024/1689) — how we comply

• Art. 50 transparency: every AI output ships with a visible disclaimer.
• Confidence: retrieval answers expose similarity bands so you can judge them.
• Human override: authors can edit any AI-generated field directly in Studio.
• Explainability: answers cite source pages; no black-box outputs.

Models used

We currently use Google Gemini (text, embeddings) and Mistral OCR via the Lovable AI Gateway. We route through EU endpoints whenever the provider offers them.

Reporting a problem

If you spot an AI output that's wrong or harmful, use the override controls to correct it and reach out via the address on your account.