After confirmation for each send, 1 selected image is sent through Supabase to Anthropic (United States) to read schedules, tasks and other content. We do not access your entire photo library or camera. Images are resized and converted to JPEG on your device and use your AI allowance. ToSche does not save the image itself in chat history or storage, but saves extracted content and replies as part of the conversation. Temporary device copies are removed after processing. Anthropic retention, processing and international transfer terms are described in Privacy Policy Sections 4-2 and 4-5. Do not attach content you have no right to share, including other people’s personal information.
This page briefly explains what information is sent where, and what is stored, when you use ToSche's AI features (the AI assistant, goal-setting chats, and voice input). The formal rules are in the Privacy Policy (Sections 4-2 to 4-5) and the Terms of Service (Sections 15-2 to 15-5). Privacy Policy: https://todo-phoneapp.pages.dev/legal/tosche/en/privacy.html
Before any information is first sent to the AI features, we show you on screen what is sent and where (to businesses in a foreign country), and ask for your consent. You can withdraw consent at any time from "Settings" → "AI & Privacy". Once withdrawn, you cannot use the AI features or voice input until you consent again, and no information is sent. Even without consent, all features other than AI remain available. The AI features can be used by people aged 13 or over. When you start using them, we ask for your age group, and people aged 13 to 17 confirm that they have a parent's or guardian's consent (Section 11).
• Anthropic, PBC (United States of America): to write the AI's replies and carry out the operations you ask for (such as adding events or tasks). When the AI looks something up on the web, the search terms are also sent through Anthropic to the search service it uses. • Google LLC (United States of America): to turn voice input into text. Information about providing data to businesses in a foreign country (that country's system and the measures each company takes) is in Section 4-5 of the Privacy Policy.
• What you type to the AI, and the recent conversation • Events, tasks, deadline tasks, goals, routines, and other data the AI reads in order to answer (only when needed, and only as much as needed) • AI memory and the "AI shared memo" (sent with each conversation; if you turn off "Use memory", memory is not sent) • In goal-setting chats, the goal being discussed and its steps, tasks linked to the goal, and summaries of earlier chats • Your device's time zone and the App's language setting • Trends in how you use the App (such as the usual color for each frequently used event title; not stored) • When the AI learns automatically from a conversation (this feature is being rolled out gradually), part of the message that triggered it and the beginning of the previous reply • If you use voice input, your audio recording (sent to Google) • If you are aged 13 to 17, that fact (sent together with safety instructions for minors)
• Passwords, sign-in information, and payment information • Your device's location, contacts, photos, and camera • App data the AI did not read in order to answer (except what Section 3 says is sent with each conversation)
• Anthropic has stated publicly that it does not use data received via its API to train AI models, and that it deletes the inputs and outputs it receives, as a rule, within 30 days. However, they may be kept longer, for example if a violation of its usage policies is suspected. • Google's contract terms with us provide that it processes audio data in line with our instructions. • These are what each company has published or agreed in its contract; we have not verified how each company operates.
• AI conversations (what you typed and the AI's replies, a record of the actions the AI took, the sources of its answers, and your ratings of replies) are stored on ToSche's servers (the database is in Japan, Tokyo region) until you delete the conversation or your account. • AI memory and the AI shared memo are stored until you delete them or your account. • Audio and the text transcribed from it are not stored on ToSche's servers. • ToSche does not use your conversations or data to train AI models. Our team reads ratings of replies to improve the quality of the AI's replies.
The AI saves things that help it plan your events and tasks (such as your school or work, class or shift times, what your words mean, and your preferences) as "AI memory". You can view, correct, and delete them in "Settings" → "What AI remembers". Especially sensitive information such as medical history or religion is saved only when you clearly ask and tap the consent button. However, it is not saved for people aged 13 to 17 or for people whose App Store or other store country is the United States. Passwords, card numbers, and similar information are never saved. If you turn off "Learn automatically from chats and app use", the AI does not learn automatically from conversations.
AI answers and AI memory may contain errors. You can check and correct events and deadline tasks the AI added on their own screens. For important decisions about health, law, money, and similar matters, please consult a professional (Section 15-3 of the Terms of Service).
If you connect ToSche to an external AI service such as Claude or ChatGPT, that service reads and writes your ToSche data within the scope you allowed. How that service handles data is governed by its own terms. You can see and remove connections in "Settings" → "Connected apps" (Section 4-4 of the Privacy Policy).
You can delete conversations from the history on the AI screen or from "Settings" → "Delete chat history". You can delete AI memory from "What AI remembers". If you have any questions, please contact support@synthera.jp.
• The AI is a program, not a person. Its answers can be wrong. For important matters, talk to a trusted adult, such as a parent, guardian, or teacher. • Do not enter personal information such as your address, phone number, school name, or passwords into the AI. • If you are feeling down or have worries about your body or mind, talk to an adult you trust or a professional helpline instead of the AI (in Japan, the 24-hour Children's SOS Dial 0120-0-78310; in the United States, call or text 988). • For people aged 13 to 17, we give the AI safety instructions for minors. If a problematic answer still appears, please let us know from that answer's menu ("More" → "Report this response"). • For parents and guardians: consent to the AI features can be withdrawn in "Settings" → "AI & Privacy". For questions about your child's use, please contact support@synthera.jp.