Your research is not advertising inventory.

Information the research system processes

When you use ChatCatharsis, the research body—including archives, prompts, candidate responses, Disagreement Maps, Round 2 movement evidence, syntheses, decisions, receipts, and new accepted source originals—is stored in browser-managed storage on the device first. Device-local storage can be removed if you clear site data, reset the browser profile, or lose the device, so important work should also be exported or optionally backed up.

A signed-in account is not required for device-local research custody. Signing in is used for shared live-model access, teams, provider connections, billing or entitlement state, compact account ledger events, usage accounting, and optional cloud backup. If you enable cloud backup, the current workspace and selected original files may also be stored in the authenticated Supabase workspace. Turning backup off stops new automatic full-workspace and source-file backups; it does not by itself delete an older backup.

Research Intake browser extension

The extension acts only after you press Preview and then affirm the disclosure and press Send governed draft. It may handle selected or visible page text, or one PDF, image, or research file that you explicitly choose. File handling includes the filename, media type, size, locally extracted text when available, original bytes, and a SHA-256 byte digest; the extension does not receive the local file path.

Pending handoffs stay in extension-local storage until ChatCatharsis acknowledges the matching intake. Accepted originals are capped at 5 MB and stored in the device vault after byte and envelope digests verify. When a signed-in user has explicitly enabled cloud backup, the original may also be copied to a private authenticated storage bucket. The extension does not collect cookies, passwords, form values, hidden page data, or browsing history; it does not intercept network traffic, monitor browsing in the background, use analytics, or contain AI-provider credentials.

AI providers

Device-local storage does not mean live inference happens only on the device. For a live run, the relevant prompt and instructions are sent through the ChatCatharsis server to the provider declared in the execution record. Candidate responses needed to produce the Disagreement Map or an explicitly requested synthesis are also sent to the declared compiler provider. Provider-specific privacy and retention terms apply to material sent for inference.

Shared inference uses only provider credentials explicitly approved for that shared role. Groq is the default shared provider when configured. Operator credentials for OpenAI, Anthropic, xAI, DeepSeek, or other paid-capable providers are not automatic fallbacks. Google Gemini shared use is separately gated and remains disabled unless the operator has explicitly confirmed that the configured Google account and data-use terms are appropriate for shared research traffic.

A workspace owner may enter a provider key in the dedicated connection form. It is submitted once over TLS, verified with that provider, and encrypted in Supabase Vault. The key is not stored in the device research vault, archives, exports, receipts, or client-readable database fields, and it is never returned after submission. Workspace members may use an owner-connected provider without seeing the key.

Teams

Team workspaces store membership, verified invitation email, role, entitlement, and audit events needed to operate shared account capabilities. Device-local research does not become shared merely because a user joins a team. ChatVaultAI desktop vault contents remain local and are not uploaded by the ChatCatharsis team feature.

Use and sharing

ChatCatharsis does not sell research content, use it for behavioral advertising, or require advertising telemetry. Information is processed to deliver the requested research workflow, preserve the records or backups you choose, secure the service, account for provider use, and comply with law.

Accounts and payments

Supabase processes account authentication and stores account, entitlement, team, provider-connection, usage, and compact ledger data associated with your account. When cloud backup is enabled it also stores the backed-up research workspace and selected source originals. Stripe processes subscription checkout, payment details, invoices, and the customer billing portal. ChatCatharsis stores Stripe customer, subscription, price, status, and period references needed to grant access; it does not store full card numbers or card security codes.

Your choices

You control when capture, file selection, transfer, generation, evidence preservation, synthesis, ratification, export, and supersession occur. You may keep research on the device, request persistent browser storage when supported, enable or disable optional cloud backup, export your workspace and receipts, download a locally held or backed-up original, remove the extension, or clear local site data through your browser.

Security and retention

ChatCatharsis uses browser-origin isolation for the device vault, server-side provider credentials, authenticated identity for account services, bounded provider requests, digest verification, and event chaining. Local browser storage is not presented as an independent cryptographic vault and can be lost with site data. Billing webhooks are signature-verified and recorded for idempotent entitlement processing. No system can promise absolute security. Account, billing, usage, compact ledger, and any cloud-backup records are retained while needed to provide the service, comply with law, resolve disputes, and enforce agreements.

Contact

Privacy and security inquiries may be sent through the contact method published by the ChatCatharsis site owner. Do not include passwords, provider keys, or sensitive research content in a vulnerability report.