Privacy and data handling
Information about the current implementation and selected retention rules. Publication status and the effective date are shown above. Financial-record retention grounds and processor and international-processing details include unresolved items.
1. Operator and contact
Operated by ZeroGravity Lab (제로그래비티랩). Send requests to access, correct, delete or restrict processing of personal data to support@talk2publish.com. We will explain the minimum identity verification and the scope of processing your request.
Payment, refund and consumer-case evidence
We separately retain minimum provider-verified amounts, currency, tax, available processing times and service periods, together with the offer and versions of terms and refund information captured at checkout. Each version separately records publication status and its effective date. Preserving a review draft does not establish acceptance of finalized terms. Necessary consumer complaint or dispute references and actual closure dates are tracked separately.
Unknown historical amounts or terms are not replaced with current values. Card details and full correspondence are not added to these records. Retention starting points and applicable tax and accounting classifications remain under review. Financial records have a read-only retention review; automatic financial deletion is disabled.
2. What information is used?
| Information | Purpose and storage |
|---|---|
| Email, user identifier and authentication information | Firebase account authentication. Social account passwords are not stored in Talk2Publish profiles. |
| Name, role, audience, tone, private boundaries, language and account lists | Personalization and profile restoration. Stored in the account profile; some settings also remain in the browser. |
| Interview audio, topic, conversation and reviewed brief | Sent to AI services for interviews and draft generation. Conversations and drafts may be saved in your browser. |
| Retry copies of results, request identifiers, hashes and counts | Recovery and duplicate-credit prevention. Result copies can be retrieved for 24 hours, separately from usage records. |
| Connected account identifiers, names, authorization tokens and expiry | Access to approved accounts. Tokens are encrypted on the server; disconnecting deletes the stored connection information. |
| Approved publishing content, destinations, status and external post identifiers | Content is sent to your selected platform. Direct-posting and newsletter job records store content hashes, lengths, destinations and status. |
| Payment and subscription identifiers, allowance and refund status | Entitlements and payment verification. Full card numbers are not stored in service profiles. |
| Request counts, models, token usage and voice duration | Usage limits and cost tracking. Daily and operation-level usage records are currently retained. |
| Support email, messages, attachments and replies | Service support and privacy requests. Emails sent to support@talk2publish.com are processed in Zoho Mail. |
3. Connected processing services
Locations shown are verified from current settings and official information. Storage location alone does not establish every country where a provider processes data. Recipient entities and international-processing details, including transfer timing, methods, purposes and retention, must be finalized before public launch.
Paddle processes buyer information under its own policy as the payment seller. Content sent to a platform you select is also subject to that platform’s policy.
| Service | Processing scope | Verified storage or processing location |
|---|---|---|
| Google Firebase | Authentication and account profiles | Authentication and current profile storage: United States |
| OpenAI API | Voice interviews and text generation | Actual project region and contract details under review |
| Cloudflare, Inc. | Site delivery, server processing and service records | US-based provider with a global network. D1 was created with an Asia-Pacific location hint; this is not a country-residency guarantee |
| OpenAI ChatGPT Sites (previous hosting) | Restricted rollback copy of pre-migration server data | Writes to the old site are stopped. Managed storage countries, recovery periods and final deletion remain unverified |
| Paddle | Checkout, subscriptions, invoices, payment methods and refunds | Entities and countries applicable to this seller account under review |
| Zoho Mail | Receiving, storing and answering support and privacy emails | Current mail account data center: United States |
| Your connected publishing platforms | Authorized account access and approved publishing content | Subject to the selected platform’s processing terms |
4. Retention and deletion: current behavior
Deleting a record in the app does not immediately remove it from Cloudflare D1 recovery history, which covers up to 30 days on the Paid plan. Operator-created encrypted migration backups and the restricted copy on the previous host are not automatically deleted on that schedule. Their final retention and deletion rules require separate confirmation.
- You keep browser conversations and drafts; they are not automatically erased after 24 hours. Use workspace reset, account deletion or clear browser site data. Remove copies on other devices or site addresses, downloads and external posts yourself.
- Server retry copies cannot be retrieved after 24 hours. Hourly scheduled work and background cleanup on site requests erase expired result content and temporary authorization states in batches. Cleanup intervals, failures and backlog may delay database deletion after retrieval ends. Hosting-provider recovery copies follow separate retention terms.
- Completed AI usage records and operational records of published jobs are cleaned up after 90 days. Unfinished or uncertain jobs, duplicate-credit prevention, payment and refund records, and evidence explicitly held for a dispute are excluded from this ordinary cleanup.
- Disconnecting removes that stored server connection. It does not also delete publishing job records or platform posts.
- After identity verification, account deletion removes the profile and sign-in account, connection tokens, interview, generation, publishing and usage records. Operational evidence held for a dispute or legal request, payment, subscription and refund identifiers and states, paid periods and used counts, and minimal deletion progress are retained separately as needed. These records may still be personal data. Separate retention grounds and periods and provider retention conditions must be settled before public launch.
| Data | Retention | Deletion and exceptions |
|---|---|---|
| Browser conversations and drafts | Until you delete them | Reset, account deletion or browser site-data removal; other devices and downloads require separate deletion |
| Server generation recovery | Retrievable for up to 24 hours | Then hourly/request-driven cleanup; access expiry and physical deletion can differ |
| Completed AI and successful publishing records | 90 days after completion | Pending, uncertain, held and financial records are excluded from general cleanup |
| Cloudflare D1 recovery history | Up to 30 days on Paid | Deleted app data can remain in recovery history; manual encrypted backups are separate |
| Closed support email | One year after inquiry closure | Manually reviewed by the operator; Zoho trash and recovery periods are separate |
| Payment, refund, dispute and tax evidence | Record-specific grounds and starting dates under review | Excluded from general cleanup and account deletion; no automated financial purge |
Deletion records and recovery safeguards
To prevent deleted data from reappearing after an older backup is restored, a separate Cloudflare D1 database records only the account identifier, deletion phase and request, completion and update timestamps. It does not record email addresses, drafts, social tokens or payment instruments.
Recovery requires the latest independent deletion records to be reapplied. Account access and social connection callbacks also check this record. It is not removed before relevant recovery copies have expired or been erased. Final retention and disposal rules must still be aligned with the manual backups and previous hosting copies.
Provider deletion and recovery copies
Firebase states that removing authentication information from live and backup systems can take up to 180 days after the account-deletion request. Authentication security IP logs are kept for a few weeks. These periods do not establish a common deletion period for profiles or every service record.
The current site runs in the operator’s own Cloudflare Workers and D1 account; Cloudflare, Inc. is the hosting processor. Paid D1 Time Travel supports recovery to a point within the previous 30 days. This differs from the service’s 24-hour result-retrieval period. Database deletion does not immediately erase every provider recovery copy. Prior account-deletion and erasure requests are reapplied before restored data is used.
The former OpenAI Sites data and an encrypted backup are retained with restricted access for migration verification. Final cleanup of the former hosting data and its provider recovery-copy conditions are being checked separately.
Support emails and account deletion
The selected retention period for inquiry emails, replies and necessary attachments is one year after the inquiry closes. The operator tracks closure dates and due dates and removes expired records manually. Emails are not automatically deleted based on inquiry closure dates. Minimum evidence needed for ongoing disputes or legal retention is managed separately.
On October 5, 2026, we verified a 30-day Spam/Trash cleanup setting for the support mailbox. It periodically removes messages in these folders based on receipt age; it does not guarantee 30 days in Trash after deletion. Zoho states that recovery copies remain for an additional 30 days after Trash removal. One year after closure is the operator’s cleanup rule, not immediate erasure of every provider copy.
Deleting your service account does not automatically erase inquiry emails in Zoho Mail. Contact support to request their deletion. We explain any limits where a request, dispute or legal retention is ongoing.
5. AI provider retention and sharing
Requests disable separate text-response and voice-session storage. We do not use stored voice recordings for download or continuation. Storage and training sharing are separate settings. OpenAI states that API data is not used for training by default unless sharing is enabled.
On October 5, 2026, we verified disabled input/output sharing and the service’s organization/project connection. Public AI availability is shown separately at the top of this page. Remaining processing-country and contract checks are required before public launch.
Abuse-monitoring logs may remain up to 30 days and some temporary caches up to 24 hours, with exceptions. API call logging was verified as enabled per call. Disabling storage or sharing does not immediately erase past data or every provider record or establish approved Zero Data Retention. Provider retention is separate from the service’s 24-hour recovery copy.
6. Your choices
Voice interviews require browser microphone permission. They cannot proceed if permission is withheld. Social connections are optional; draft creation does not require them.
Turns marked private are excluded from subsequent text-generation inputs. Audio already spoken may have been processed by the interview AI, and your profile’s private-boundary instructions are themselves sent to AI. The private flag does not prevent all AI transmission in advance. Do not input sensitive or confidential information belonging to others.
7. Privacy requests and objections
Send your registration email, reply email, request type and scope to support@talk2publish.com. Do not send passwords, verification codes, API keys or full identity-document copies. We will explain the minimum verification needed to confirm you or an authorized representative.
After reviewing identity verification and scope, we respond within the applicable statutory time limits. If access is deferred or part of a request must be restricted, we explain the affected records, reasons and how to object. You can request reconsideration in the same inquiry or use external privacy advice and redress channels.
‘Download current conversation and drafts’ in account settings exports content from the current browser. It is not a complete personal-data export including profile, usage and payment evidence. Request other records separately if needed.
Account deletion, removal of support emails and removal of external platform posts have different scopes. Minimum records required by law and provider recovery copies may not be erased at the same time. We distinguish completed actions, separately retained records and grounds, and any other responsible provider.
8. Privacy contact
Joonho Jang (장준호), the representative of ZeroGravity Lab, serves as the privacy officer and handles privacy inquiries.
Send access, correction, deletion or processing-restriction requests, privacy questions and complaints to support@talk2publish.com. We will explain the necessary identity verification and request scope.
9. Before public launch
Legal bases, data-specific retention and deletion, processor and international-processing details, and the effective date must be finalized for actual operation. This page does not replace that review or establish a final privacy policy.