Picture Dictionary Kids

Privacy Policy

Privacy contact: privacy@skillworks.dev

Product Scope

Picture Dictionary v1 is a child-directed English picture dictionary. v1 applies strict child-protective defaults to all users.

v1 does not include login, account creation, ads, third-party analytics, tracking, social features, user uploads, external web browsing, payments or in-app purchases, or cross-device sync.

Initial release excludes China mainland.

Learning History

Learning history is local-only on the device. The app stores searched words, lookup dates, Calendar history, Words list, Day list, and local lookup counts in the app container.

Clear Learning History deletes this device's word history and learning calendar. Shared dictionary entries and shared dictionary pictures remain available for all learners.

Deleting the app or clearing app data may also remove local learning history. v1 does not provide cloud recovery or cross-device sync for learning history.

Backend Processing

Searched words or short phrases are sent to the backend so Picture Dictionary can validate input, reject unsafe or invalid input, return learner-friendly dictionary content, generate or retrieve safe shared dictionary pictures, and maintain shared dictionary image quality.

The backend stores shared dictionary assets only: normalized word or short phrase, learner definition, example sentence, IPA, part of speech, primary sense, other common meanings, canonical image URL, image status, image job metadata, and aggregate image-quality feedback.

The backend does not persist raw search queries as child-specific records. The backend does not store a child's Calendar, Words list, Day list, or learning path in the cloud for v1.

Device token, anonymous app identity, IP address, and related request metadata are used only for internal operations such as security, abuse prevention, rate limiting, cost control, troubleshooting, and infrastructure reliability. They are not used to build child profiles.

AI Images And Feedback

Dictionary pictures are AI-generated on the backend, checked for safety before becoming shared canonical images, and identified in the app as AI-generated learning pictures.

If an AI-generated picture is still being created, the app shows the word information first and displays a pending picture state. If picture generation fails, the word information remains visible and the app may offer a safe Try again action.

The app includes simple image-quality feedback controls: Like and Not a good picture. This feedback is used as aggregate shared dictionary image-quality feedback. It is not used for advertising, tracking, analytics, child profiling, or a child learning path.

Images with enough negative aggregate feedback may be marked for manual review as replacement candidates. A single dislike does not automatically regenerate or replace an image.

OpenAI And Pronunciation

OpenAI is used only server-side. The Apple app never calls OpenAI directly and does not contain an OpenAI API key.

OpenAI receives only normalized dictionary words or short phrases and safe backend-generated dictionary/image prompts. OpenAI does not receive device identifiers, install ids, app tokens, IP addresses, local learning history, child profiles, raw rejected personal-information-like input, raw rejected unsafe input, or full server logs.

Pronunciation uses Apple on-device AVSpeechSynthesizer. v1 does not send searched words to a third-party text-to-speech provider.

Service Providers

Picture Dictionary v1 uses Google Cloud for backend hosting, database, secrets, logging, and monitoring; Cloudinary for approved canonical dictionary image storage and delivery; OpenAI for server-side dictionary assistance, image generation, and image safety validation; and Apple for App Store distribution and on-device pronunciation features.

OpenAI, Google Cloud, and Cloudinary are reviewed as service providers/processors and are expected to provide the same or equivalent level of data protection expected for a child-directed app.

Data Not Collected In V1

v1 does not collect child name, email, birthday, school, precise location, photos, contacts, microphone audio, or free-form uploads.

v1 does not sell data and does not use data for third-party advertising, developer advertising or marketing, third-party analytics, or tracking.

Image Safety Takedown

Parents, guardians, or reviewers can report an unsafe, inaccurate, or inappropriate dictionary image by contacting privacy@skillworks.dev or support@skillworks.dev.

The operator reviews reported images, removes or replaces confirmed unsafe images, and preserves only the operational records needed to investigate and prevent recurrence.

Regional Notes

EU/UK Article 27 representative requirements must be evaluated before broad EU/UK public launch. China mainland launch is excluded for v1.