Privacy Policy

All core functionality runs locally on your machine. The following data is created and stored only on your device — it is never transmitted to us or any third party, with two exceptions: the anonymized usage statistics described in Section 4 (which include individual dictionary-form words but never these records themselves), and the optional Playto Cloud described in Section 6, which is off by default and sends only recognized text when you explicitly turn it on:

• Your personal dictionary (saved words, sentences, translations)
• Learning progress (quiz results, spaced repetition state, streaks)
• Application settings and game profiles
• Screen captures used for text recognition (stored only on your device — for example recent capture history and album images you keep — and never transmitted)

We have no access to this data. If you uninstall Playto, this data is deleted with it.

Playto downloads open-source AI models (for text recognition, translation, and text-to-speech) from external hosting services such as Hugging Face. These are one-way downloads — model files are retrieved to your device, but no personal data is sent to the hosting provider.

Standard connection metadata (such as your IP address) is visible to the hosting provider during the download, as with any file downloaded from the internet. No additional information is transmitted.

Playto provides tools to create session share cards that may include game screenshots. When you choose to share a card (e.g., to X/Twitter or save to your device), the image is sent directly from your device to the platform you selected. Playto does not upload shared images to our servers.

The privacy policy of the platform you share to (e.g., X/Twitter) governs how they handle the shared content.

Game screenshots may be subject to the game publisher's content usage policies. It is your responsibility to verify that sharing screenshots of a specific game complies with that game's terms. Playto provides the sharing mechanism but does not grant rights to third-party game content.

To improve Playto, it sends us limited, anonymized usage statistics. This is on by default and can be turned off at any time in Settings → Advanced → Telemetry.

What we collect:

• App version, edition (free, Steam, or Pro), platform, and UI language.
• Feature, mode, and capture-method usage (e.g. which translation mode a session used).
• Language settings and language pairs (e.g. Japanese→English).
• Setup progress, days since install, and launch counts.
• Error codes when something fails (the code only, no logs or text).
• Vocabulary engagement — when you look up, save, or review a word: the dictionary form of that word with its part of speech, a coarse difficulty signal, the review result, a coarse non-identifying skill bucket, and the game title and Steam App ID. These records carry a month-level timestamp; setup and error events use day-level dates.

What we never collect:

• Screen captures, screenshots, or the text you read in games — we send individual dictionary-form words, never sentences or passages.
• Translated sentences, free text, or personal files.
• Anything that identifies you: no name, account, email, or Steam identity. IP addresses are not stored or linked to your data.

Identifiers: A randomly generated installation identifier (not linked to your Steam account or any account) is used only for aggregate setup and milestone statistics — such as first launch, setup completion, model readiness, and first successful translation. These records mark that a step was reached; they carry no text, image, or file content. All other statistics, including error codes and vocabulary records, use a token that changes daily and cannot be linked across days.

How it's used: Data is anonymous and aggregated across many players. Vocabulary records build a dataset of game-vocabulary difficulty for language learners; we publish only aggregated statistics (e.g. "this word is hard for learners of game X"), never individual records and never any game's raw vocabulary list. We may publish or license aggregated statistics. We never sell your personal information. Where a legal basis is required, we rely on our legitimate interest in improving Playto.

Your control: Turn it off anytime in Settings → Advanced → Telemetry; turning it off stops all sending and clears anything pending. Settings shows the full list of what is sent.

Game publishers: to have a game excluded from the aggregated dataset, contact [email protected].

Contribution license: You grant Playto a worldwide, royalty-free, perpetual, irrevocable, sublicensable license to use the anonymous learning signals derived from your usage (e.g. that a word was looked up, saved, or graded) to compute, publish, or license aggregated, anonymized statistics. You represent that providing these signals does not require rights you do not hold. Playto does not collect or publish verbatim game text.

If you contact support from inside Playto, a diagnostic report is sent only when you choose to send it. It can include application logs (with recognized and translated text masked out), system information such as GPU and CPU model, your settings, profile names (masked by default), and an email address if you provide one. Reports are used only to investigate and respond to your request.

Playto can optionally translate — and generate dictionary entries — using our cloud service instead of a local AI model. It is off by default and is used only after you turn it on in the app. You can switch back to local processing at any time.

What is sent: only the text recognized in the screen areas you selected, and, if you enable cloud dictionary generation, the saved words being defined. Screen captures and images are never sent.

Where it goes: the text travels over an encrypted connection to our relay service, which passes it to a third-party AI provider that returns the translation or definition. Our relay does not store this text — it is not written to our databases or logs. We only use providers whose terms do not permit using this content to train their models.

Current providers: OpenAI (United States). For users in regions where OpenAI does not provide service, requests are routed instead to Alibaba Cloud Model Studio (Singapore), under terms that likewise do not permit training on this content. We may change or add providers; this section is updated when we do.

Service account: to grant and meter the free trial we verify through Steam that you own Playto, and keep a service account keyed to your Steam ID containing an access credential (stored as a hash), your remaining balance, and daily aggregate counts of requests and tokens. No text is stored in this record. It is separate from, and not linked to, the installation identifier used for the Section 4 statistics.

• Steam (Valve Corporation) — Playto is distributed through Steam. Valve collects data according to their own privacy policy (e.g., playtime, purchase history). We do not receive personal data from Valve.

Playto is not directed at children under the age of 13. We do not knowingly collect any data from children under 13.

We may update this Privacy Policy as Playto develops. Significant changes will be communicated through the application or our Steam store page. The "Last updated" date at the top reflects the most recent revision.

If you have questions about this Privacy Policy, contact us at:

Email: [email protected]