Privacy Policy

This Privacy Policy describes how the browser extension "VLM Camera" ("the Extension") handles information. The same terms apply to shared-PC administrators and users, whether or not cloud features (Google Drive backup, browser notifications, etc.) are used. Even if you do not use cloud features, please understand how data is handled if those options are enabled.

1. Nature of the Extension
   The Extension is a free tool that analyzes camera frames on-device using vision-language models (VLMs), OCR, and related components to generate descriptions and assist with condition-based detection and notifications. Analysis and detection results are probabilistic estimates, not definitive judgments. The Extension is provided as-is, with no warranty of operation, accuracy, or availability.
2. Camera and Images
   A camera is used for analysis. After each capture, the camera is generally released.
   The Extension does not store or transmit images or video as history records, Google Drive backups, browser notifications, or uploads to the web service. Even when cloud features are enabled, image payloads are not included in backups.
   However, to let you review the latest result in the popup or similar UI, the Extension may temporarily keep the most recent capture frame in local extension storage, overwriting the previous one. This is not intended as evidence retention or long-term image archiving. It is overwritten by newer captures and is not included in history lists or Drive backups.
3. Data Kept Inside the Extension
   Main information kept inside the Extension includes:
   ・Settings (model / runtime, prompts, schedule, notification and Drive options, language, generation and detection parameters, etc.)
   ・Administrator password-related data (to protect settings, history, and the popup; a skip option may exist)
   ・Cached inference models and OCR language packs (on device)
   ・Local AI response and prompt history in IndexedDB (responses and prompts are always stored encrypted and decrypted only for display)
   ・Optional saved translations, model-load / inference progress, and the latest preview frame (as above)
   Without cloud features, this data is generally kept only in the browser storage on your device.
4. Model and Translation Downloads
   Models required for inference are downloaded to and cached on your device (which involves network requests to third-party model hosts). If you use Chrome's built-in translation features, the browser may download language packs. The Extension does not operate its own cloud translation server that receives your text for translation.
5. Optional Cloud Features (device registration / paid tier)
   If you register the device on yourcut.net and enable cloud features, the following may apply:
   ・Text backup of AI responses: backed up to your own Google Drive. Images are not included. Drive backups are plaintext by default. If you turn on encryption, later backups are ciphertext and cannot be read by opening Drive files directly. Encryption keys needed for decryption are not stored on Google Drive; they are kept on the device and, when linked, per device on the backup service provider's (web service) servers. Past plaintext uploads are not automatically re-encrypted. Managing, sharing, deleting, and retention of Google Drive data is the responsibility of the person who configured backup.
   ・Browser notifications (e.g. FCM): registration data and tokens needed to send notifications may be synced with the web service and notification infrastructure. Notification payloads do not include images.
   ・Device IDs and link tokens: stored on the web service to identify the device and maintain linkage.
   Transfer of backup payloads is performed by the Extension or the user's browser (client-side) connecting directly to Google's services (Google Drive). The service provider neither stores this backup payload nor relays the communication.
   There is no fee for the Extension itself. Charges relate to maintaining web services such as backup and notification linkage; see the web service terms and privacy policy for details.
6. Community Templates
   When cloud features are enabled, you may publish model and prompt / detection-related settings to the community, or search and apply others' templates. Published content is intended to be usable by anyone. Do not put personal information in names, descriptions, or tags. Notification settings, Drive settings, and device tokens are not included in templates. You are responsible for what you publish and apply.
7. Administrator and User Responsibilities
   Signage, consent, and legal compliance for camera use in shops, workplaces, or shared spaces are the responsibility of the deployer (administrator). Risks of operating Drive backups in plaintext are also the responsibility of the person who configured them. Shared-PC users should understand that an administrator operates the Extension and, when enabled, cloud features on that device. Do not put personal information in identifiers, template names, or prompts.
8. Retention and Legal Compliance
   Local history limits and similar settings can be changed. If laws in your country or region prohibit or limit long-term log retention, adjust settings accordingly.
9. Changes
   This policy may change without prior notice. Important changes will be announced in the Extension settings or similar.