LingoGrasp is a camera app that recognizes, translates, and reads aloud text and objects in the world around you. This policy explains what data the app handles, why, and the choices you have. The short version: you can use LingoGrasp without an account, in which case your scans stay on your device; if you sign in and enable cloud sync, your scans and photos are stored in our cloud so you can restore them across devices — and you can delete everything, in the app, at any time.
Data we process
| Data | When | Why |
|---|---|---|
| Photos you capture or pick, and the text/objects recognized in them | Every scan | Recognition, translation, study notes, and text-to-speech. Recognition runs on your phone where it can (Google ML Kit, no network). When you use a cloud feature, the photo or the recognized text is sent to our servers (AWS) and from there to the AI processors named under Where data lives; it is used only to produce your result and our servers do not keep the photo. To make repeat scans fast and affordable, the results — recognized text, object labels, translations, readings, study glosses, and speech audio — are cached on our servers, keyed by a fingerprint of the content rather than by your identity, for 90 days (study glosses: 180 days). |
| Your scan library (photos, recognized text, translations, titles) | Only when signed in with cloud sync on | Backup and cross-device restore. Stored in AWS (us-west-2), scoped to your account. |
| Email address | If you create an account (email, Google, or Apple sign-in) | Account identity, sign-in, and verification emails. Managed by AWS Cognito. |
| Purchase and subscription status | If you subscribe | Unlocking Pro features. Payments are handled by Apple; our subscription processor RevenueCat receives a pseudonymous app user ID and purchase receipts — never your card details. |
| Usage counters (e.g. daily scan counts) | While using cloud features | Fair-use rate limiting and abuse prevention. Counted per account, or per anonymous install ID when you are not signed in; each counter expires 24 hours after your last use. |
| Your device's network (IP) address | When you use cloud features without being signed in | Abuse prevention only: a per-address cap on how often an unsigned-in device can verify a Pro purchase, fetch Japanese readings, or repair a stuck account. The address is stored as a usage counter, kept for 24 hours after the last request, and is visible to our operators in an internal admin console during that time. It is not linked to your scans or your account. |
| Crash reports and diagnostic events | On errors | Fixing bugs. Processed by Sentry. Personal data and scan content are scrubbed before sending; when you are signed in, a crash report carries a pseudonymous account identifier (not your email) so we can tell one affected account from another. Reports are not used for advertising or profiling. |
What we do not do
- No ads, no selling of data, no sharing with data brokers.
- No use of your photos or text to train machine-learning models.
- No tracking across other companies' apps or websites.
- No microphone access — the app speaks to you (text-to-speech) but never records you.
Where data lives, and who processes it
Cloud data is stored with Amazon Web Services in the United States (us-west-2). The companies that process data for us ("processors"), and what each one receives:
- Amazon Web Services (United States) — account sign-in (Cognito), storage of your scan library and photos, our servers and caches, and Amazon Textract, which recognizes text in photos written in Latin scripts (English, Spanish, French, German, and similar). Textract receives the photo, not your identity.
- OpenRouter (OpenRouter, Inc., United States) — the routing service our servers use to reach the AI models below. For cloud text recognition in other scripts, object detection, translation, and study notes, our servers send OpenRouter the photo or the recognized text together with our instructions, and OpenRouter forwards it to the model host. Our request tells OpenRouter which hosts it may use (listed next) and sets its "data collection: deny" option, which restricts routing to hosts whose published policy is not to store or train on prompts. OpenRouter records the size and cost of each request for billing.
- Google (United States) — runs the Gemini 3.8 Flash and Gemini 3.1 Flash Lite models that do our cloud text recognition and translation and write study notes (word glosses and explanations) from recognized text. Reached only through Google's own AI endpoints (Google AI Studio / Vertex AI) via OpenRouter. Google receives the photo or text being processed. Separately, if you choose "Sign in with Google", Google provides your email address and name to our sign-in.
- Claude Haiku 5.5 — Anthropic's model that does our object detection and stands in when our other models are unavailable. It is run for us by Google (Vertex AI) or Amazon Bedrock (Amazon Web Services, United States), reached via OpenRouter; whichever runs it receives the photo or text being processed. Anthropic itself does not receive it.
- Microsoft Azure AI Speech (United States) — generates speech audio for some languages; receives the scan text only.
- RevenueCat (United States) — subscription management; receives an anonymous app user ID and App Store receipts, never your email or card details.
- Sentry (United States) — crash reporting, as described above.
- Apple — payments and "Sign in with Apple" (if you choose it).
On-device recognition (Google ML Kit, running entirely on your phone) sends nothing anywhere. No AI processor receives your name, email, or account identifier; they see the content being processed and our server's network address. Every connection between your phone, our servers, and these processors is encrypted (TLS). We will update this list here before we add or change a processor.
Retention and deletion
Guest-mode scans exist only on your device — deleting the app deletes them. Server-side result caches, described above, are content-keyed, not linked to you, and expire within 90 days (study glosses within 180 days). Cloud data is kept until you delete it: individual scans can be deleted in the app (this also removes the photo from cloud storage), and Account → Delete Account permanently removes your account, your entire scan library, your photos, and your usage records. Things that expire on their own rather than being deleted with the account: usage and network-address counters (24 hours after last use), records of requests we declined for exceeding a limit (35 days), and internal cost-accounting entries that carry a pseudonymous caller key but no content (90 days). Crash reports are kept by Sentry for its standard retention period. Deleting your account does not cancel an App Store subscription — manage that in your Apple account settings.
Your rights
Depending on where you live (e.g. GDPR in the EU/UK, CCPA in California), you may have rights to access, correct, export, or erase your personal data. The in-app deletion flow covers erasure; for anything else, contact us at lingograsp.com/support and we will respond within 30 days. We do not "sell" or "share" personal information as those terms are defined by the CCPA. All of our processors are in the United States; where data is transferred outside your region, we rely on standard contractual clauses and, where a processor is certified under it, the EU–U.S. Data Privacy Framework.
Children
LingoGrasp is for readers of all ages — including children learning to read, pronounce, and discover words they have never seen before, whether in their own language or a new one. Children can use the app without an account, and guest-mode scans stay on the device as described above. Creating an account or turning on cloud sync does involve personal data (an email address and a cloud-stored scan library), so for a child under 13 (or the equivalent minimum age in your jurisdiction), a parent or legal guardian must create and manage the account and accept this policy on the child's behalf. We do not knowingly collect personal data from a child under that age without a parent's consent — if you believe we have, contact us and we will delete it.
Changes
We will post any changes to this policy here and update the date above. Material changes will be highlighted in the app.
Contact
LingoGrasp is operated by Panya Ou (United States), who is the data controller for the personal data described in this policy. Reach us at support@lingograsp.com.