Privacy & data use

1. Content you choose to share

Choosing “Verify with DeltaMind” creates a local draft from selected text. It does not automatically submit the claim. You review and explicitly submit selected or pasted text, optional source context and follow-up questions. The extension does not automatically scan full pages, capture screenshots, read page cookies or collect browsing history in the background.

Fresh selections include a sanitized source URL by default, with the URL and sharing control visible before submission. Uncheck the control or clear the URL to exclude it. Query strings and fragments are removed; URLs with embedded credentials are rejected. The remaining path can still reveal private information. Existing drafts keep their sharing choice, and manual drafts do not need a source URL.

Website features also process what you deliberately submit, including uploaded media, extracted text and metadata when media-upload features are available. The extension itself does not upload images or full pages. Public creation and processing availability can be disabled independently of sign-in.

2. Accounts and authentication

Google Firebase Authentication handles email/password and optional Google sign-in, verification and password-reset messages. Account data can include email, Firebase user ID, verification status and provider profile information. Passwords go directly to Firebase through its SDK, not to DeltaMind’s investigation API.

The API verifies a Firebase ID token and derives an account ownership identifier. Verified email is required for private records. Google access tokens used for the extension’s sign-in exchange are not sent to the investigation API. Authentication identifiers are not added to model prompts unless you include them in submitted content yourself.

3. Processing and recipients

• DeltaMind and Google Cloud: the extension uses a fixed DeltaMind HTTPS gateway. The main service and database run in Google Cloud’s US east region. The service retains submitted context, results, evidence, conversations, certificates, provider-response/audit records and request reservations for history, continuation, integrity and abuse/cost controls.
• Google/Firebase: authentication and account security use Google services. Firebase can process security metadata such as IP addresses and user agents under its service practices.
• OpenAI and search/source services: investigation and follow-up inputs and supporting context go to the configured OpenAI API. Web-search tools and evidence websites can receive derived searches or retrieval requests. Processing can occur outside the main service’s region.

Hosting and operational logs can contain IP addresses, user agents, request paths/timing, status codes and error information. These are separate from the content-free usage analytics described below.

Provider requests use store=false; this does not guarantee zero provider retention, zero logging or a particular training arrangement. Provider settings and terms apply separately. Read OpenAI’s data controls and Firebase’s privacy information.

4. Private records and public history

Private investigations and discussions are checked against their account owner on the server. This is access control, not end-to-end encryption: operators and processing providers can handle data needed to operate the service. Earlier shared-archive records remain public. Signing in does not claim, migrate or privatize them. Do not treat a shared archive link as a private link.

Usage reports require an explicitly authorized verified administrator account. Reports contain aggregates, not account pseudonyms, individual claims, URLs or record identifiers. Signing in alone does not grant administrator access.

5. Local browser state

The extension uses chrome.storage.session for short-lived Firebase ID tokens/expiry, owner-bound drafts, pending request keys and investigation references. Its Firebase SDK state is memory-only; it does not use extension sync/local storage for those sessions or drafts. Website sign-in is also memory-only. Website drafts can remain in that tab’s browser session across sign-in/reload.

Closing the panel is not sign-out. Signing out, reloading/updating or disabling the extension, or closing the browser can clear its session. This does not delete an account or server records, revoke Google cookies/grants, or cancel accepted work. Pause stops local waiting, not backend processing. After extension session loss, reopen saved records on the website after signing in; the extension has no server-history browser.

6. Optional first-party usage counts

Collection is disabled in this release. If enabled later, signed-in workspaces display a notice and send a content-free workspace open. The server deduplicates account/client opens into 30-minute UTC buckets and records minimal operation admissions, outcomes and error categories. Seven- and thirty-day reports count accounts, not necessarily unique people. This feature adds no anonymous visitor tracker, fingerprint, tracking cookie or third-party analytics SDK.

Analytics rows contain pseudonymous account keys, bounded client/kind labels, internal correlation handles and timestamps. They exclude raw claims, questions, answers, URLs, emails, tokens, IPs and user agents. Pseudonyms are account-linked data, not a claim of anonymity. Existing investigation data and infrastructure logs are separate. Turning collection off does not instantly delete retained data.

7. Retention and data requests

• Application records: private and legacy public investigations, discussions, stored media-analysis results, reservations and existing audits have no automatic expiry or self-service deletion API. Certificate integrity protections block ordinary modification/deletion. Deleting a Firebase account does not automatically delete these records.
• Analytics: rows have a 90-day cleanup threshold, measured from the visit bucket or operation admission. Cleanup runs on analytics writes/report reads or an explicit operator purge. No scheduled purge is installed; exact physical expiry at 90 days on an idle service is not promised. First-activity timestamp metadata can remain without account-level rows.
• Infrastructure: standard Google Cloud logs are configured for 30-day retention; required platform audit logs for 400 days. Database backups retain seven backups, a count rather than a guaranteed deletion period. Backup and provider copies can outlive primary records.

Contact the address above to request access, correction, account handling or deletion, or to ask about current retention. We may need to verify ownership. Requests need manual review of integrity, security, backup, provider and legal constraints; immediate deletion or a fixed completion deadline is not promised. Never send a password or access token with a request. This notice does not limit rights you may have under applicable law.

8. Use limits and policy changes

We limit data use to the disclosed DeltaMind features, support you authorize, security and necessary operations. Data is not used for personalized advertising, data brokerage or lending decisions. Human access is limited to authorized support, necessary security/abuse handling or legal obligations. Transfers are limited to the described service processing and necessary legal/security purposes.

The use of information received from Google APIs will adhere to the Chrome Web Store User Data Policy, including the Limited Use requirements.

Material changes to purposes, recipients or collection will be reflected here and in applicable product disclosures before that changed processing is enabled. This notice does not claim absolute security, complete anonymity or a legal-compliance certification. Service and provider limits can pause processing while saved records remain readable.