Skip to main content

Privacy and security

Your manuscript stays yours

How Publication Intelligence handles manuscript and account data, including current safeguards and limitations.

Privacy commitment

Your manuscript stays yours

Publication Intelligence processes manuscript-derived content to prepare, index, review, store, troubleshoot, secure, support, and operate the requested service. AI-provider handling varies by endpoint, feature, tier, account setting, agreement, safety event, and law; we do not make a provider-wide no-training or zero-retention promise.

For complete legal disclosures, read the Privacy Notice and Terms of Service.

Private by design

The current application is designed to scope projects to their authenticated owner, require authenticated manuscript downloads, use revocable server-side sessions and HTTP-only cookies, and store manuscript files in cloud storage configured to prevent public access. These are checked-in controls, not a guarantee about every live setting or every authorized provider or personnel access.

Project deletion

When you delete a project, Publication Intelligence first marks it deleted, cancels active work, and removes it from normal account views. Cleanup of active database rows, manuscript objects, and tracked provider state then proceeds and may be retried. Acceptance of the request does not mean every active byte, recovery copy, provider copy, or retained legal or business record has already been erased.

Recovery systems and AI providers follow separate schedules and exceptions. Publication Intelligence requests deletion only for tracked stored OpenAI Responses; Anthropic content follows its applicable retention schedule, and Gemini batch deletion is not currently implemented. We do not promise immediate deletion or deletion of every provider, safety, legal-compliance, backup, or recovery copy.

Authorized access and confidentiality limits

Private does not mean that no other system or person can ever process a manuscript. Authorized personnel and service providers may access or process content when reasonably necessary for requested support, security or abuse response, recovery, legal compliance, or a disclosure you authorize. No online service can guarantee absolute confidentiality or security.

AI processing and provider retention

Depending on the selected workflow and production configuration, Publication Intelligence may send manuscript-derived prompts to commercial APIs operated by OpenAI, Anthropic, and Google’s Gemini Developer API. The inspected application does not use provider file uploads or tuning, but stored OpenAI background Responses and Gemini inline batches create provider-side state. Anthropic does not offer Publication Intelligence per-request deletion for paid API content.

Provider notices current as of September 24, 2026 describe different training, retention, abuse-monitoring, deletion, and account-control rules. Read the current notices from Gemini, Anthropic, and OpenAI. Google’s storage recovery behavior is separately documented in its soft-delete documentation. Public provider notices do not establish which terms or settings apply to Publication Intelligence’s live accounts.

Current limits and legal review

Publication Intelligence does not claim a security certification, fixed data residency, verified qualified-human review, or review-gated exports. Live provider controls, production access settings, recovery settings, and legal terms require separate verification; the product remains early access and public registration is disabled.

Use the private workspace for project files

Create an account before adding a project. Do not send manuscript text, index entries, or other confidential publication content through ordinary inquiry channels.

Use an approved account, not manuscript email

Public registration is disabled during early access. Do not send final PDF page proofs through ordinary inquiry channels.

Early access