This notice applies when you choose to use AutoLog AI features — document recognition from a photo (AI Scan) or voice input (AI Dictate). AI features are optional; you can enter data manually. Cloud processing must not start until you accept the current notice.
AI Dictate is offered only when your device supports speech recognition. If voice input is unavailable, you can still use AI Scan or enter data manually.
AutoLog provides a redaction tool. After you review and redact a document image, AutoLog sends the redacted/edited image itself to Google services (Firebase / Gemini) for multimodal field recognition. AI Scan does not send OCR text instead of the image and does not run recognition only on your device. The unedited original is not intended to be sent to the AI service.
When you choose voice input, AutoLog uses the device’s speech recognition service (often Google) to turn speech into text. AutoLog does not send an audio file to Gemini or to AutoLog’s servers.
After speech is recognized, AutoLog sends the transcript text (and technical metadata below) to Google services (Firebase / Gemini) for structured field suggestions — the same JSON contract as AI Scan.
The request may include the document or record type, account and request identifiers, an app/device integrity token, subscription tier, and quota usage. It does not include your Google password or payment-card details.
Photo AI Scan and AI Dictate share one monthly AI quota per account: Pro — 80 and Business — 600 AI requests per UTC calendar month. Each successful Gemini request counts toward this limit. Speech recognition alone does not consume quota if Gemini is not called.
AI Scan: Before continuing, remove names, addresses, signatures, document numbers, payment details, faces, and any other information you do not want to disclose. You are responsible for personal data that remains visible.
AI Dictate: You are responsible for what you say and for personal or confidential information in the transcript. Do not dictate another person's data without a lawful reason and permission.
Google processes the edited image or transcript text to return structured field suggestions. Processing and any service-side retention are governed by the applicable Google Privacy Policy and Firebase / Gemini terms. Audio handled by the device's speech recognition service is governed by that service's terms (often Google).
AutoLog does not use submitted images or transcripts to train its own model and does not intentionally persist them in its own database. Results are shown for your review and stored in AutoLog only if you save them. Firebase stores the user/account identifier, Pro or Business tier, monthly quota usage, request reservations, and abuse-prevention records needed to operate and secure AI features.
AI features are optional. You can:
If you use AI, review every suggested field before saving (see also the Terms of Use).
If this notice changes materially, AutoLog increments the consent version and asks you to accept again before another cloud AI request. Removing the app clears the locally stored acceptance record; cloud account and quota records are handled as described in the Privacy Policy.