Privacy Policy

Last updated: October 12, 2026 · Version v1.2-2026-10-12 · Applies to stayclassy.ai and the Socrates app (socrates.stayclassy.ai)

Version notice. This dated update adds the initial mobile-release disclosures below and preserves the existing website provisions. Mobile-specific eligibility, storage, billing, and cleanup provisions are identified separately. Read the previously published version v1.0-2026-09.

Classy AI ("we", "us") operates the Socrates tutoring service. This policy describes what information we collect, why we collect it, who processes it on our behalf, how long we keep it, and the controls you have. Two commitments up front: we do not sell your personal information, and Classy AI does not train AI models on your data. Section 6 explains exactly how far that second commitment reaches.

1. What this policy covers

Our public website at stayclassy.ai and the Socrates application at socrates.stayclassy.ai. The two behave differently in one important way — the website uses analytics and the application does not — so this policy says which is which throughout. Where an institution has a separate written agreement with us, that agreement governs the deployment it covers.

2. Information we collect

• Account information. Your email address, an optional display name, a securely hashed password (we never store the password itself), your chosen role (student or educator), and subscription status.
• Tutoring conversations. The full text of your chat sessions with the tutor — your messages and the tutor's replies.
• Uploaded documents. Files you upload, such as lecture notes or textbook chapters — the original file, the text extracted from it, and the filename.
• Quiz and learning records. Quiz questions, your answers and scores, per-section proficiency estimates, study streaks, badges, and class enrollment records.
• Model-inferred learning signals. Classifications our AI systems make about your learning activity — for example, which cognitive level a quiz result reflects — including short model-written rationales that may paraphrase your work.
• Logs and security records. An audit record of significant account actions, including IP addresses and timestamps, kept for security and academic-records integrity; standard server logs; and request counters used to enforce rate limits.
• Billing identifiers. A customer record is created with our payment processor when an account is registered, including for students whose access is sponsored by an institution; a subscription record exists only if you buy a personal plan. Card details are collected and held by the payment processor, never by us.
• Roster information from educators. If an educator rosters a class, they may provide student email addresses in order to invite those students.
• Contact and chat-widget messages. If you use the contact form or the chat widget on our website, the email address and message you submit.

3. Why we use it

• Providing Socratic tutoring grounded in course materials.
• Adapting quiz difficulty and review recommendations to your measured performance.
• Giving educators insight into their own classes: educators see progress, concept-level evidence, and short excerpts of a student's own quiz work or of a message showing an attempt — never a student's full chat transcript. Downloadable topic reports list what a class asked about by date, document and section, with AI-written summaries; they don't include students' names, emails or IDs.
• Account security, fraud prevention, and audit integrity.
• Billing and subscription management.
• Service email: welcome messages, password resets, and class invitations sent on an educator's behalf. We do not send marketing email.
• Error monitoring, and page-view analytics on our public website only (section 9).

4. Service providers

We use a small number of service providers to run the service. Each receives only what its function requires, and each is engaged to process that information for us rather than for itself. By category:

• Hosting and storage providers — run the application and hold service data at rest and in transit, including the rate-limit counters described above.
• AI processing providers — produce tutoring replies, quiz questions, document captions, and learning classifications from conversation text, course-material text, and in some cases page images.
• Document-search providers — convert course-material text and study questions into the numerical representations used to search your materials.
• Payment processors — create the customer record described above and handle subscriptions and card details.
• Transactional email providers — deliver welcome messages, password resets, and class invitations.
• Error-monitoring and security providers — receive error reports, configured to exclude personal information.
• Website analytics providers — receive page-view events from our public website only, never from the application.

Our website and these legal pages also load web fonts from a third-party font service, and the contact form and chat widget use a third-party bot-detection service. Like any web request, those requests carry your IP address and browser information to the provider serving them.

Institutions evaluating the service can ask us for the current subprocessor register — the named providers, their purposes, and their locations — under a written agreement.

5. AI processing, in plain language

When you chat with the tutor, generate a quiz, or upload course material, that text — and, for some documents, images of the pages — is sent to our AI processing and document-search providers so that a reply, a quiz, or a searchable index can be produced. Three things are worth being precise about:

• We do not attach your identity to those requests. Your name, email address, and account identifier are not included in what we send.
• We cannot promise your personal information never reaches a provider. The request carries what you wrote or uploaded. If you type personal information into a message, or upload a document that contains it, it is sent along with everything else. Treat the tutor accordingly.
• Providers handle that content under their own contractual terms, which is the subject of the next section.

6. Selling, advertising, and model training

• We do not sell your personal information, and we do not share it for advertising.
• Classy AI does not train AI models on your data. We do not build or fine-tune models, and we run no training of any kind on conversations, documents, quizzes, or learning records.
• What a provider may do with content we send it is governed by that provider's own terms. We are compiling a written record of each provider's training and retention commitments. Until that record is complete, this policy makes no promise about provider-side training beyond what is written here, and you should not infer one. Institutions can request our current written position on any provider.

7. How long we keep information, and what deletion covers

We keep your information while your account is active. We have not yet adopted fixed retention periods for each category; when we do, we will state them here.

When you delete your account, we delete in a single database transaction: your conversations and messages, your uploaded documents and everything derived from them — including study questions generated from them, which are deleted with the document — your quiz attempts and learning records, model-inferred learning signals, your class enrollments, class invitations addressed to your current email address, and password-reset tokens. After the database transaction, we attempt to remove your uploaded original files, cancel your subscription, and delete your payment-provider customer record. Failed cleanup remains queued for retry rather than blocking account deletion; payment-provider cleanup and its retention exceptions are described in section 15.

Documents other people studied. Deleting a document — or deleting your account, which deletes your documents — removes the file, its text and its search index, so no one can study it any more. Conversations other people had about that document are theirs, so they are kept: each stays in that person's history, marked as being about a deleted file, and can be read but not continued. Your own conversations about your own documents are deleted with them. An instructor who uploaded a file that should never have been shared can choose a separate action, "Delete file and students' conversations", which also permanently deletes those conversations and everything derived from them; it requires typing the document's name and can be used only by the document's owner or the class's staff.

Some things do not go, and you should know which:

• Audit records. Records of significant account actions, including IP addresses, are currently kept indefinitely for security and academic-records integrity. When you delete your account, your account reference is removed from them and any email address in them is replaced with an irreversible token.
• Invitations to a former address. A class invitation sent to an email address you no longer use cannot be linked to your account, so account deletion does not reach it.
• Backups. Copies can remain in routine system backups after deletion.
• Provider copies. Content already sent to a service provider is subject to that provider's retention terms (section 6).
• Conversations about a deleted document. As described above, other people's conversations about a document you deleted remain in their accounts unless the file was removed with "Delete file and students' conversations". They are removed when that person deletes them or their account.

8. Your controls

• Export. You can download your chat history and your learning data — document metadata, quiz attempts, proficiency, engagement, badges, and classes — as JSON from Settings at any time.
• Correction. You can change your display name and email address in Settings. For anything you cannot correct yourself, write to us at the address in section 14.
• Deletion. You can delete your account yourself from Settings, with the coverage and the exceptions described in section 7.
• Institutional requests. Where an institution sponsors accounts, export and deletion beyond these self-service controls are carried out by us on the institution's written request.

9. Cookies, browser storage, and browser signals

The Socrates application sets no cookies and contains no analytics code: nothing about your tutoring, quizzes, documents, or account is sent to an analytics provider. It stores your sign-in token and your app preferences in your browser's own local storage so that you stay signed in and your settings persist.

Our public website uses a third-party analytics service for page-view measurement, which sets its own cookies; it also loads web fonts, and a bot-detection service on the contact surfaces. That website holds no student data. There is no cookie-consent banner on it today, so the controls available to you are your browser's own cookie settings.

We do not currently respond to "Do Not Track" or Global Privacy Control signals sent by your browser.

10. Who the service is for

The service is designed for higher education. It is not directed to children under 13, and we do not offer accounts to children under 13.

11. Institutional use and FERPA

When Classy AI is deployed by a school, college, or university, student learning records created in that deployment can constitute education records under FERPA. We support institutions in meeting their FERPA obligations under a written agreement — including contractual limits on use and disclosure, and the export and deletion controls described in sections 7 and 8. We hold no certification and no institutional approval, and nothing here is a claim of compliance on your institution's behalf. Institutions can contact us to put a data-processing agreement in place.

12. Security

Passwords are stored as salted hashes and never in readable form. Traffic between your browser and the service is encrypted in transit. Educator dashboards are scoped so that an educator can reach only their own classes. Significant account actions are recorded in an audit log. We hold no security certification and have not been audited against any security standard, and no system is perfectly secure.

13. Changes to this policy

We post changes on this page and update the version and the "last updated" date above. We do not currently send individual notices when this policy changes, and there is no notice inside the application, so this page is the place to check.

14. Contact

Privacy questions, access and correction requests, and anything else about this policy: support@stayclassy.ai.

15. Initial mobile release — October 10, 2026

This policy also covers the Classy AI iOS and Android applications operated by Classy AI, Inc. Initial mobile availability is limited to the United States and adults aged 18 or older. Mobile users affirm eligibility before study access. We do not collect a date of birth for this confirmation; this is self-attestation, not identity-verified age. Institutional agreements govern the deployments they cover.

The mobile applications use existing account or institution-provided access. They do not offer purchases or direct users to external purchasing. Existing website subscriptions remain managed through the website and Stripe.

Deletion for the mobile release. You can request deletion from Account settings after confirming your current password. Account deletion removes or deidentifies owned account data and invalidates account access. Original-file removal is queued for retry if immediate cleanup fails. Payment-provider cancellation and customer cleanup are also queued for retry. Account deletion may complete while provider cleanup remains pending. Only necessary payment-provider identifiers and retry metadata are retained for this purpose; each identifier is removed after successful cleanup. Persistent failures are escalated for operator attention. If billing continues after deletion, contact support@stayclassy.ai.

Shared-service cleanup. The durable cleanup behavior above also applies to website account deletion through the shared service. The existing exceptions concerning institutional records and other people's conversations remain applicable.

Retention exceptions. Minimized security and audit records, independent billing and provider records, institutional records, and backups may remain. Their treatment follows the existing exceptions in section 7, actual legal or institutional obligations, and applicable provider terms. We have not adopted new fixed backup, log, audit, or provider retention periods through this update. In particular, no 30-day backup or 90-day log deadline is promised. No reference here implies that an additional retention arrangement has been approved.

Native storage and safety. Native authentication tokens use platform-protected device storage. Preferences and account-scoped adult and AI permission markers are stored locally. Server-side AI permission receipts and submitted content reports retain necessary account associations for authorization and safety handling. AI processing requires the normal permission-review interface; users can report an individual tutor response. Native usage analytics remain disabled in the release configuration. The existing provider-processing and training limitations in sections 4–6 continue to apply.