Check AI origin
The answers of our AI chat assistant and our AI project advisor on identic.pro are generated by AI (Google Gemini via Vertex AI Search). We mark every answer according to Art. 50(2) of the EU AI Act: visibly with the note “AI-generated answer”, machine-readably in the API response (headers AI-Generated and AI-Disclosure) and in the HTML (attribute data-ai-generated), and with an invisible marking inside the text itself that usually survives copying.
Here you can check whether a text carries this marking. The marking states source, model and time and is signed: if the signature is valid, the text is word for word one of our answers. If it is invalid, at least part of it comes from one of our answers, but the text was changed.
Limits: the marking consists of invisible Unicode characters. Programs that remove such characters, or retyping the text, delete it. It is not a watermark in the choice of words as model providers can embed — that is technically not possible when generating through a third-party service. The pasted text is only processed for the check and is neither stored nor logged.
Machine access: POST /api/ai-detection with {"text": "…"} returns verdict (marked-unchanged, marked-modified, unmarked) and, per marking, source, model, time and signature validity.