ScienceGuru Edu Privacy Policy

1. Who we are

AutoTrust AI Pte. Ltd., Singapore, is the controller of the personal data described here. Where your institution deploys ScienceGuru Edu for its members, the institution's administrators also have access as described in Section 4; the institution is responsible for its own use.

2. What we collect

Account data: username, email address (when you sign in with Google, GitHub or an email code, or an administrator sets one), the identity provider's account id, invitation chain (who invited you), sign-in times and addresses, the desktop version and operating system.

• The Research Record (see Terms, Section 5): every model request and reply including system instructions, tools offered and retries; every tool call with its inputs and complete outputs, including shell commands and their output; the before-and-after content of files the assistant edits; snapshots of your project's source tree and environment manifest (dependency versions, hardware) at the start, at key steps and at the end; experiment artifacts (scripts, data files, figures, logs, models) in the workspace; conversation titles and project directory names; which models and providers you used and when.
• Usage and billing data for account models: requests, tokens and spend per model per day, as reported by the gateway.
• Diagnostics: application logs and error reports, with secrets removed.

3. What we do not collect

API keys, OAuth access and refresh tokens, and passwords for providers you connect yourself. They are stored only on your computer; the application removes Authorization-type headers from every record before upload, and requests to your providers go directly from your computer to them.

Your password for ScienceGuru Edu is stored only as a one-way hash.

Content from applications and folders you did not open in ScienceGuru Edu. The application records the workspaces you work in, not your whole computer.

4. Why we use it and who can see it

We use the data to: provide and secure the Service; train, fine-tune and evaluate AI models and agents; study how research is done with AI; build datasets, benchmarks and publications (de-identified where practicable); detect abuse; and meet legal obligations.

• AutoTrust staff and contractors with a need to know, under confidentiality obligations.
• Administrators of your institution's deployment: per-account totals of what has been retained, usage, sign-in activity and the audit trail of actions on your account. They do not see the contents of prompts or files through the console.
• Service providers that host the data for us (Amazon Web Services, Singapore and the United States regions) and the model providers that process requests for account models under agreements that forbid training on that traffic.
• Authorities, where required by law or to protect rights, safety or the Service.
• A successor, if AutoTrust is acquired or merges, under the same commitments.

5. Legal bases

Where the law requires one: performance of the contract with you (providing the Service); your consent, given when you accept the Terms in the application, for the collection and use of the Research Record for model development and research; our legitimate interests in securing and improving the Service; and compliance with legal obligations. You may withdraw consent by declining the Terms, which ends your use of the Service; withdrawal does not affect processing already done.

6. Retention

The Research Record is retained for as long as it is useful for the purposes above, which may be indefinitely. Account data is kept for the life of the account and up to three years after. Audit logs are kept for seven years. Backups follow the same schedule with a delay.

7. Your rights

Depending on where you live you may have rights to access, correct, export or delete your personal data, to object to or restrict processing, and to complain to a supervisory authority. Write to privacy@autotrust.ai; we will respond within the time the law allows, usually 30 days.

Deletion has limits: material already used to train a model cannot be removed from that model; de-identified datasets and publications are not withdrawn; we keep what the law requires us to keep. We will delete the identifiable Research Record and account data on a valid request to the extent these limits allow.

8. International transfers

Data is stored and processed in Singapore and the United States and may be accessed from other countries where AutoTrust or its providers operate. Where required, we use contractual safeguards for such transfers.

9. Security

Data in transit is encrypted (TLS); data at rest is encrypted; access is role-based and logged; the archive upload is authenticated per account and verified by content hash. No system is perfectly secure; tell us at security@autotrust.ai if you find a problem.

10. Children

The Service is not for anyone under 18. We do not knowingly collect data from children; if we learn we have, we delete it.

11. Changes

We may update this Policy. The version date changes, and for material changes the application asks you to accept again before continuing.

12. Contact

AutoTrust AI Pte. Ltd., Singapore — privacy@autotrust.ai.