Privacy Policy

Last Updated: 2026年10月9日

Blave ("we", "our", or "us") is committed to protecting your privacy. This Privacy Policy explains what data we collect, how we use it, and your rights regarding your personal information when using the Blave mobile application and its related services.

1. Information We Collect

We collect the following types of data to provide you with a complete and secure experience:

• Account Information: Your email address and encrypted login credentials.
• Exchange API Data: If you choose to connect your exchange API, we may access your account details, trading history, and positions.
• Blave Agent Instance Data: Strategy execution status, positions, order records, and system and execution logs, used for display, anomaly notifications, and troubleshooting.
• Device Information: Device type, operating system, IP address, and app version.
• Usage Data: Your interactions with the app, preferences, and behavior analytics.
• Cookies and Analytics Identifiers: cookies and similar technologies store your sign-in session, language preference, and an anonymous analytics identifier. See Section 8.
• Blave for Desktop Usage Events and Current Settings: when you use Blave for Desktop, the app sends a limited set of usage events and, after you sign in, tells us three current settings; when you download a library strategy without signing in, we record which strategy and this installation's identifier. See Section 9.
• Contact and Report Data: when you contact us or report content through the channels on the Contact Us page (email, Discord), the email address or Discord username and the message content you provide, used to reply to and handle your message or report.
• Server Access Logs: when anyone visits this site (including public research pages viewed without signing in) or connects to our servers through Blave for Desktop, our servers record the IP address, time, requested URL, and browser information as part of routine security operations, and keep them for no more than 14 days.

2. How We Use Your Information

Your data is used for the following purposes:

• To provide core features like charting, portfolio insights, and trading analytics.
• To improve app performance and user experience.
• To send important system updates and notifications (you may opt out of marketing messages).
• To email reminders about your progress to your account address after you sign in to Blave for Desktop (see Section 9.1; you can unsubscribe at any time).
• To ensure account and platform security.
• To generate de-identified and aggregated insights that may be used for internal analytics or offered to users as premium features.
• To reply to your messages and handle reports about published content.

3. Sharing of Information

We do not sell your personal data to third parties in an identifiable form. Data is only shared under the following conditions:

• When required by law or government authorities.
• With trusted technical partners (e.g., cloud service providers), under strict confidentiality obligations.
• When you explicitly choose to share content (e.g., screenshots or portfolio summaries).
• Aggregated and de-identified data may be used internally or as part of features offered to users or partners. See Section 4 for more details.
• With third-party analytics providers (currently Google Analytics 4), which process site usage data on our behalf. See Section 8.
• With our email delivery provider (currently MailerLite), which sends email on our behalf. We share the data it needs to send: your email address, your IP address at sign-in and the country derived from it, your last sign-in time, your subscription plan, whether you have connected an exchange API, your card status, and which reminder group you are placed in. MailerLite's handling of that data is governed by its own privacy policy (mailerlite.com/legal/privacy-policy).
• When you choose to use Blave's AI in Blave for Desktop, the content of your conversations is sent to the provider of the model you select (currently Anthropic and DeepSeek) for processing. See Section 9.2.
• When you publish a Blave Agent report. If you publish a performance report, the account assets, positions, and profit-and-loss figures in it become public; if you choose to be credited, your display name (including any referral code in it) becomes public too.
• When you contact us through Discord, Discord's handling of that data is governed by its own privacy policy.

4. Use of Aggregated and Anonymized Data

We may use de-identified and aggregated data derived from user-linked exchange APIs to generate statistical insights and market indicators. These insights:

• Do not contain any personal identifiers.
• Cannot be linked back to individual accounts.
• May be made available to other users as part of premium features.

You can opt out of contributing to such analytics by contacting us or managing your privacy settings in the app.

5. Data Retention and Deletion

We retain your data only for as long as necessary to fulfill its intended purpose. You may request to delete your account and associated data at any time by contacting us.

• Snapshots and publication records of reports that are or were published are kept for 180 days after account deletion, under Section 8.8 of the Terms of Service.
• Messages and report data are deleted within a reasonable time after they have been handled.
• Blave for Desktop usage events are kept for 25 months from receipt. See Section 9.1.
• The current settings Blave for Desktop tells us after you sign in, and records of desktop reminder emails (which kind, which language, and the delivery result), are kept for 90 days. See Section 9.1.
• Records of Blave for Desktop library downloads made without signing in are kept for 25 months from creation; after that, only each strategy's cumulative install count is kept, without installation identifiers. See Section 9.1.
• AI usage records and credit transaction records are kept for the period required by law.

6. Your Rights

You have the right to:

• Access, correct, or update your personal data.
• Request account and data deletion.
• Revoke API access at any time.
• Opt out of receiving marketing communications.

7. Data Security

We implement appropriate technical and organizational safeguards to protect your information, including encryption and access control.

8. Cookies and Third-Party Analytics

We use cookies and similar technologies (such as local storage) to keep you signed in, remember your language and display preferences, and measure how the site is used.

• Essential: sign-in session, language, and display settings. Blocking these prevents you from signing in.
• Analytics: Google Analytics 4, provided by Google LLC, records page views, referral source, approximate region derived from your IP address, and device type. Google's handling of that data is governed by its own privacy policy (policies.google.com/privacy).

We do not use advertising or cross-site tracking cookies. You can block or delete cookies in your browser settings, or install Google's opt-out add-on (tools.google.com/dlpage/gaoptout).

9. Blave for Desktop

This section applies to Blave for Desktop (macOS and Windows) installed on your computer.

The desktop app periodically connects to Blave's update source (download.blave.org) to check for a new version. That connection is served by a content delivery network, and its access logs are not covered by the retention period stated in Section 1.

9.1 Usage Events

To understand where installation, setup and everyday use break off or get stuck, and how many people are on each version, the desktop app sends us the following kinds of usage events. After you sign in, these events are also used to decide whether to send you usage reminder emails (see "Purpose" below). Each event answers only one question: which step happened (or where things got stuck), when, and on which version; the "Online" item answers whether the app was open that day. Each event's name and the option values it can carry are fixed lists written into the app's code; the complete list is public in the desktop app's source code (shell/telemetry.js in github.com/Blave-TW/blave-agent). When the desktop app is updated, events or option values may be added or removed within the categories below; if we add a new category, or change the scope of what usage events "never include" below, we will revise this Privacy Policy.

• Installation, launch and online: "First open" is sent once, the first time this installation starts. "Open" is sent each time the app starts; we keep one record per installation per day (UTC). "Online" is one record for each day (UTC) the app is open, starting about 10 minutes after launch; we record whether automated trading on this computer is running at that moment (paper trading counts) — only yes or no, not whether it is paper or a real account, which exchange, or which strategy.
• Connecting an AI and signing in: when you choose which AI to use, and the first time this installation receives a reply from the AI (sent once), we record which AI it is (Blave's AI, Claude Code, Codex, or your own API key). When you successfully sign in to your Blave account in the desktop app, we record one event. When a connection does not complete, we record only the category of the cause: sign-in to Claude Code or Codex did not complete, the tool could no longer be found when you clicked connect, Blave sign-in failed or was cancelled, neither Claude Code nor Codex is usable on this computer, or your own API key did not pass verification (the key was not accepted, your balance with the provider is too low, the provider could not be reached, or another reason). For each of Claude Code and Codex that the check finds unusable on this computer, we also record which reason category applies (for example not found, timed out, or not signed in).
• Milestones: the first time a strategy finishes a backtest on this installation (sent once); when you start order placement and it succeeds, in which case we record whether it is paper trading or a real account, not which exchange; and when you confirm in the desktop app that a strategy is to be sent to the cloud and the request has been handed to the AI agent. In addition, when each strategy is created, when it first finishes a backtest, and when it is first assigned an amount for automated order placement, we record its strategy type (A, B or C) and market category (for example crypto, TAIEX futures or US stocks), not the strategy's name, code, symbols or amounts.
• Failed conversation turns: we record only the category of the cause, which is one of the following: Blave's AI could not be charged (no card or not enough credit); authorization expired or the trial allowance is locked; requests were rate-limited; the AI tool you chose could not be found; the usage limit that the desktop app sets for a single turn was reached while you were using your own API key; or another reason.
• Card, top-up and cloud plan: whether a card or top-up prompt appeared and whether you clicked it (distinguishing the check when the workspace opens from a turn that was blocked, and no card from not enough credit), and, after you clicked, whether on returning to the app your account was ready, still had no card, or still lacked credit; and the result of starting a cloud plan from the desktop app (success, no card, insufficient balance, or error).
• Strategy library and finding ideas: when a strategy library strategy cannot be used, we record the category of the cause (signed out, no card, insufficient balance, account status could not be confirmed, the cloud host is stopped, or Blave's AI is selected without a card or enough credit). When you use a library strategy on this computer, we record whether that strategy needs Blave data (needed, not needed, or not labeled), not which strategy it is. When you ask the AI agent to search the web for strategy ideas, we record which entry point the request was sent from (the welcome page, the strategy library header, or the library's empty list), not the text you entered.
• Updates and installation: the stage at which an automatic update failed (checking for a new version, downloading, verifying the signature, installing, or other); and the result of installing the engine and strategy packages, on first install or when more packages are added later (for example after an update): finished, or failed because of a network problem, a timeout, or another reason. If the optional US stock data packages fail to install, we record the same categories for them.
• Feature use: when you open a tab or use a feature, or when a feature's prompt or result appears, we record the name of that feature (for example the "Parameter scan" tab, "Pick strategies", report sharing, strategy export, which screen a card or top-up button was clicked from, or whether a chat attachment was pasted or picked from your files), never the content inside it. Some names by themselves show which integration or which kind of built-in content you are using (for example the setup steps for Capital Futures or President Futures, an export format for charting software such as TradingView, or whether a built-in morning report covers Taiwan stocks or crypto).

Apart from the three events noted above as sent once, we keep one record per day (UTC) for each installation, each kind of event, and each of its option values. None of these events records the text of error messages or any other content.

Each event carries only: an installation identifier (Installation ID), the event name, one option value for that event (if any; at most one per event), the app version, the operating system name and version, your system locale, the time on your computer, and the time we received it.

Usage events never include: the content of your conversations with the AI, the text of error messages, strategy code, strategy names, traded symbols, amounts, positions, balances, keys, or file paths. Events have no free-text field; our server stores only the listed fields and discards everything else.

• How you are identified: the installation identifier is a random code generated the first time the desktop app starts. It is stored on your computer and does not change; it is not derived from your hardware or your account. After you sign in to your Blave account in the desktop app, subsequent events carry your sign-in credential so that we can associate this installation with your account. While you are signed out, events are not associated with any account.
• Current settings after sign-in: after you sign in to your Blave account in the desktop app, each time the app checks your account status with us (for example at launch, at sign-in, on each conversation turn, or when you switch usage events or change AI), it also tells us three current settings: whether usage events are on or off; which AI it is connected to (Blave AI, Claude Code, Codex, your own API key, or none yet); and whether this computer has run a backtest (only yes or no). While usage events are off, it tells us only that they are off, not the other two. While you are signed out, it tells us nothing. These settings are used only to decide whether to send a reminder email and which one (see "Purpose" below). We record an entry for each computer whenever its settings change and keep it for 90 days; decisions use only the latest entry for each signed-in computer.
• Library downloads without signing in: when you are signed out, or the desktop app's sign-in credential is no longer valid, and you download an official strategy from the library that is free and uses no Blave data, the desktop app attaches the installation identifier to the download request. We record which strategy (its library number), which installation, and when it was first downloaded, and use this only to count the strategy's installs; repeated downloads of the same strategy by the same installation are recorded once. This record is not a usage event; it refers to a public strategy in the library and contains none of your own strategies, conversations, or keys. If you later sign in to your Blave account on this computer, the installation is associated with your account (see "How you are identified" above), so this record can be matched to your account as well. While usage events are off, the download request carries no installation identifier and we record nothing; the download works as usual. To prevent bulk downloading, downloads are rate-limited by IP address; the counters expire automatically and are not otherwise kept. For retention, see Section 5.
• IP address: usage event records do not contain your IP address or browser information. As with any network request, however, connections between the desktop app and our servers appear in our server access logs (see Section 1).
• Purpose: usage events are used only for these two things. (1) To measure completion of installation and setup, where people get stuck and the category of the cause, daily presence and retention, feature usage, and version distribution, in order to improve the product, and, when you give us your installation identifier while reporting a problem, to help find its cause. (2) To send usage reminder emails: after you sign in to your Blave account in the desktop app, we use the usage events the app sends while you are signed in and the current settings described above, together with your account's card, credit, sign-in and cloud host status, to decide whether to email your account address a reminder about your progress. There are currently two: you picked Blave AI but have not added a card; or your AI is ready but you have not run a first backtest. Each kind is sent at most once every 90 days, and an installation that has never signed in never receives one. Reminders are sent by our email delivery provider (see Section 3); see "How to turn it off" below for how to stop them.
• Retention: usage events are kept for 25 months from receipt.
• How to turn it off: usage events are on by default. You can turn them off at any time under Settings › Privacy in the desktop app. Once turned off, the desktop app sends no usage events at all, including "First open" and "Open", and including any event queued but not yet sent at that moment. Turning it off does not delete records already sent. Turning it back on resumes sending immediately. If an event cannot be delivered because you are offline or the request times out, the app discards it; it does not retry or queue it for later. If an event that is sent only once or once a day is not delivered, it is sent again the next time the same situation occurs. There are two ways to stop reminder emails. (1) Use the unsubscribe link at the bottom of a reminder. Unsubscribing applies to all email sent through that delivery service, so reminders, newsletters and other marketing email all stop; verification codes and essential notices about failed charges, low balance and your cloud host are sent separately and are not affected. (2) Turn off usage events in the desktop app. If any computer signed in to your account reports that usage events are off, we send no reminders to this account until you turn them back on.
• Requesting deletion: for usage events that are not associated with an account, you can email info@blave.org with your installation identifier and ask us to delete the usage events and library download records for that installation. You can view and copy the installation identifier under Settings › General in the desktop app.
• When you delete your account: we remove the association between usage events and your account. The events themselves are kept, unlinked from any account, until the retention period ends. Desktop sign-in credentials issued to your account are revoked, and events received after deletion are not associated with that account.
• Keys: exchange API keys and bring-your-own data source keys that you set up in the desktop app are written to a workspace file on your own computer (Blave/workspace/.env in your home directory; on macOS, ~/Blave/workspace/.env) and are not sent to Blave's servers for storage. That file sits on your computer, and Blave Agent, your strategy code and the desktop app itself can read it — that is how they trade and fetch data for you. Your own API key for an AI provider (see Section 9.2) is not written to that file. It is encrypted using the encryption facility provided by your operating system and stored on this computer in the desktop app's own folder (on macOS, Keychain holds the encryption key; on Windows, the system's built-in data protection is used), and it is not sent to Blave's servers. The desktop app does not put it into the AI agent's runtime environment; it sends it directly to that provider, and only when verifying the key and when sending your conversations. When you use Blave's AI or your own API key, the results of tools the AI agent runs are sent to the model provider under Section 9.2; if you ask it to read a file that contains keys, that content falls within what is sent on that call. Two exceptions, both started by you: (1) when you send a strategy to the cloud, the data-source keys that strategy uses travel with it to your Blave Agent cloud machine, governed by Section 7 of the Terms of Service — the confirmation dialog lists what moves and what does not, and exchange API keys do not move; (2) when you connect an exchange for your cloud machine from the desktop app, the exchange API key you enter is relayed through Blave's servers to your own cloud machine and written there; the relay queue entry is removed as soon as your machine picks it up, an entry never picked up expires within 15 minutes, and Blave keeps no further copy. Exchange API connections you make on the website remain governed by "Exchange API Data" in Section 1.

9.2 Conversation Data When You Use AI

The desktop app lets you choose who provides the AI:

• Claude Code or Codex on your own computer: the desktop app runs the tool already signed in on your computer and does not use Blave's AI relay. Your conversations do not pass through Blave's servers; they are handled by that tool's provider under your own terms with it.
• Blave's AI: the content of your conversations — including the messages you enter, system prompts, the results of tools the AI agent runs, and images or files you attach — is relayed through Blave's servers to the provider of the model you select for processing, and the reply is relayed back to you. The providers are currently Anthropic and DeepSeek, depending on the model you select in the desktop app. Requests are sent under Blave's account with the provider. The request content is produced by the AI agent software and relayed as is; Blave's relay service does not add your Blave account identifier to it.
• Your own API key (DeepSeek is currently supported): you obtain a key from the model provider and paste it into the desktop app, and that provider bills you directly. The content of your conversations — including the messages you enter, system prompts, the results of tools the AI agent runs, and images or files you attach — is sent by the desktop app from your computer directly to that provider. It does not use Blave's AI relay and does not pass through Blave's servers. Requests are sent under your own account with the provider, and the content is handled by that provider under your own terms with it. Before the key is saved, the desktop app uses it to send that provider one test request that contains none of your data, to confirm that the key works. For how the key is stored, see "Keys" in Section 9.1.

When you use Blave's AI in the desktop app, Blave's servers do not keep conversation content: it is not written to any database, cache, or log file. Reply content is held in memory only while it is being relayed, in order to read usage figures, and is released when the request ends.

For billing, we record for each AI call: your account identifier, the request path, the model name, input and output token counts, the number of web searches, and the time. Your credit transaction history records the amount charged and a description containing only the model name and those counts. None of these records contain conversation text. They are kept for the period required by law.

Content received by a model provider is handled under that provider's terms and policies. How long it is kept and how it is used are determined by the provider's current documents:

• Anthropic: Commercial Terms of Service, Privacy Policy
• DeepSeek: Open Platform Terms of Service, Privacy Policy

Conversation history for the desktop app is stored on your own computer.

10. Policy Updates

We may update this Privacy Policy from time to time. The updated version will be posted here with a revised "Last Updated" date.

11. Contact Us

If you have any questions about this Privacy Policy, please contact us: info@blave.org