Privacy policy

Effective October 24, 2026

Anatomy is a 3D viewer of the human body, used inside ChatGPT and Claude. It is operated by Kushal Sokke Malleshappa (Kushal SM). This page describes what Anatomy receives and keeps.

Anonymous access and account-enabled ChatGPT connections

The published Anatomy connection works without sign-in. The separately enabled ChatGPT account connection requires you to sign in with Google before use and links new activity to your Anatomy account. Claude stays anonymous. Anonymous browser history is not retroactively linked to an account.

What Anatomy receives

When ChatGPT or Claude opens or changes the viewer, Anatomy receives the scene it asked for: structure names, the viewing angle, labels, movements and similar settings. While you use the viewer, it receives what you do in it: which structure you tapped, what you hid, faded, isolated, peeled or cut, where the camera is, how far a movement has played, quiz answers, and any place you marked with "Hurts here".

A "Hurts here" mark is a structure name and a point on the reference model, with the step or movement frame you were on. It describes a location you chose; Anatomy does not know anything else about the symptom. Treat it as health-related information you are choosing to share.

How it is used

This information is kept so the viewer can restore your view if it reloads, and so the assistant can read what you are looking at when you ask a follow-up. It is returned to ChatGPT or Claude only when it reads the viewer in that conversation.

Where it is stored and for how long

Each viewer gets a random session identifier. Its scene and your actions are stored on Cloudflare (Workers and Durable Objects) under that identifier, and deleted automatically seven days after the last change. Cloudflare also keeps short-lived request logs used to troubleshoot errors. Information may be processed outside your country.

OpenAI operates ChatGPT and Anthropic operates Claude. Each handles your conversations under its own policies. Deleting an Anatomy session does not delete anything already in your ChatGPT or Claude conversation.

Usage measurement

For about 90 days, Anatomy keeps tool names, call times, execution outcomes, software deployment versions, and a country code supplied by Cloudflare. It also keeps broad source labels when request headers identify a client, and optional owner or test labels supplied by the caller. These labels are not verified identities. The country can describe a proxy or server rather than your location. Tool usage records link to your Anatomy account after sign-in. On the signed-in connection, each tool call also carries one short sentence from the assistant describing what you asked Anatomy for, such as “studying the knee for an exam” or “understanding shoulder pain”; we keep it with your account to understand who uses Anatomy and why. It can describe a health topic you raised. The anonymous connection does not record it. Usage records contain no raw IP addresses, scene or session identifiers, raw queries, conversation text, or pain marks. They help us measure tool use and diagnose failures; automatic viewer polling is not counted as tool use.

The browser also sends a receipt after rendering a 3D anatomy frame and after your first deliberate interaction each day. Where browser storage is available, a random identifier is kept locally for up to 90 days to estimate repeat browsers. If storage is blocked, only the separate viewer instance is counted. These records keep hashed anonymous browser and scene identifiers, a random viewer-instance identifier, receipt times, broad host/platform claims and the country Cloudflare observes on the browser request. After sign-in, receipts link to your account. We also record broad feature categories, such as selecting a structure, changing layers, or using a quiz. These categories contain no structure names, anatomy questions, quiz answers or symptom text. They are retained for about 90 days. A browser identifier is not a verified person; different devices or cleared storage produce different identifiers.

Your account

For the ChatGPT connection, you sign in with Google. Claude remains available without sign-in. Anatomy stores your Google account identifier, email and its verification status, name, account creation time and last update. Connection records include the requesting client and redirect host, sign-in time and the country Cloudflare observes on that sign-in request. This is a request location, not your nationality or residence. Google identity tokens are not retained as account data, and Anatomy asks Google only for your basic profile and email. Sign-in does not give Anatomy access to your ChatGPT history or your Google account. Account records and the index of account-owned viewer identifiers remain until account deletion; usage and sign-in records are retained for about 90 days. OAuth connection grants and refresh tokens remain until revoked or expired.

What Anatomy does not do

It does not sell information, show ads or use third-party analytics, or use your information to train AI models. Temporary, secure cookies bind the sign-in and consent flow to your browser.

Contact

Questions or deletion requests: kushalsokke@gmail.com. Use the email associated with your ChatGPT sign-in. You can revoke the connection in your host app. For an account-enabled connection, ask ChatGPT to delete your Anatomy account and confirm permanent deletion; this removes your account, linked usage and account-owned viewer sessions and revokes connections. If cleanup is interrupted, access is blocked and deletion is retried by the daily cleanup. Contact us if you cannot access the connection. Copies already in host conversations and provider-controlled logs or backups are outside this deletion.

This policy may change as Anatomy changes; the effective date above will be updated.