ScriptaHubScriptaHub · working draft
AI transparency
Where ScriptaHub uses AI and what readers should expect.
1. Books and editions
ScriptaHub publishes long-form work created or revised with generative AI. A book page should identify its edition, available source formats, change history and feedback route. Complete the editorial policy for human review, factual checking, citations, corrections and responsibility: [EDITORIAL POLICY].
2. AI Librarian
The AI Librarian compares a reader’s question with book titles, descriptions, categories and discovery terms. It returns up to ten relevant paths through the catalogue. Recommendations are relative matches, not endorsements or automated decisions about a person.
3. Limits
Recommendations can miss synonyms, irony, context and expertise requirements. A high displayed match means that catalogue language overlaps with the question; it does not prove quality, truth or suitability. The percentage is relative to the top result for that question and is not a probability.
4. Reader control and feedback
The question remains on the device and in the URL fragment. Readers can inspect the matched keywords, open any result, change the question or use the separate feedback form. The form opens email and does not silently upload the question.
5. Publication transparency
For AI-generated or manipulated public-interest text, document whether human review and editorial responsibility apply and add any disclosure required by Article 50 of the EU AI Act. Creative and fictional work should still disclose its artificial origin in a way appropriate to the work. Complete the machine-readable marking and provenance approach before the relevant publication duty applies: [IMPLEMENTATION].