Data Processing Addendum

1. Parties and scope

This Data Processing Addendum (“DPA”) forms part of the agreement (the Terms of Service at https://www.infinitelibrary.ai/terms, together with any order or plan, the “Agreement”) between the customer that accepts it (“Customer”) and Infinite Library SAS, a French société par actions simplifiée with its registered office in Montrouge, just south of Paris, France (“Infinite Library”, “we”), for the Bindery studio and the Infinite Library website (the “Services”).

It applies wherever Infinite Library processes Personal Data on Customer's behalf as a processor. It is accepted online by a person authorised to bind Customer; the acceptance record (version, fingerprint, signatory, date) is emailed to the signatory and kept in Customer's account.

2. Definitions

Terms such as “Personal Data”, “processing”, “controller”, “processor”, “data subject” and “personal data breach” have the meanings given in the EU General Data Protection Regulation 2016/679 (“GDPR”). In addition:

• “Customer Content” means the manuscripts, transcripts, notes, recordings, images, prompts, instructions and other material Customer or its users bring to the Services, and the works generated from them.
• “Sub-processor” means a third party engaged by Infinite Library to process Customer Content, as listed in Annex III.
• “SCCs” means the Standard Contractual Clauses adopted by the European Commission in Decision (EU) 2021/914, Module Two (controller to processor) or Module Three (processor to processor) as applicable, and the UK International Data Transfer Addendum where UK law applies.

3. Roles

Customer is the controller of Customer Content (or a processor acting for its own controllers, in which case it warrants it has the authority to instruct Infinite Library). Infinite Library is Customer's processor for Customer Content.

For the account, billing and usage data of the individuals who sign in to the Services, Infinite Library acts as an independent controller as described in its Privacy Policy (https://www.infinitelibrary.ai/privacy). This DPA does not apply to that data.

4. Details of processing

The subject matter, duration, nature and purpose of the processing, the types of Personal Data and the categories of data subjects are described in Annex I.

5. Infinite Library's obligations as processor

Infinite Library shall:

• process Customer Content only on Customer's documented instructions — the Agreement, this DPA, and the use of the Services' features by Customer's users — unless required by Union or Member State law, in which case it will inform Customer before processing unless the law prohibits it;
• ensure that every person authorised to process Customer Content is bound by confidentiality;
• implement the technical and organisational measures in Annex II, and maintain them at a level appropriate to the risk throughout the term;
• engage Sub-processors only as set out in Section 7, and remain responsible for their performance;
• assist Customer, taking into account the nature of the processing, in responding to data-subject requests — the Services provide self-serve export and deletion so that most requests need no assistance at all;
• assist Customer with security, breach notification, data-protection impact assessments and prior consultation, taking into account the information available to Infinite Library;
• make available all information necessary to demonstrate compliance with Article 28 GDPR, and allow for and contribute to audits as set out in Section 9;
• delete or return Customer Content at the end of the Services as set out in Section 10.

6. No AI training on Customer Content

Infinite Library does not train, fine-tune or otherwise improve any machine-learning model with Customer Content, and does not permit any third party to do so on its behalf.

Customer Content is transmitted to AI Sub-processors solely to perform the function Customer's user invokes (writing, editing, translating, illustrating, narrating, transcribing) and is retained by them only for the abuse-monitoring periods stated in Annex III. Infinite Library engages each AI Sub-processor on terms under which Customer Content is not used to train the Sub-processor's models, or maintains that Sub-processor's account-level training opt-out; the live status of each is recorded, vendor by vendor, in Annex III.

Product analytics processed by Infinite Library are pseudonymous event counts and never contain the text of Customer Content.

7. Sub-processors

Customer gives general authorisation to the Sub-processors listed in Annex III, which is the live register published at the Trust Center and incorporated here by reference.

Infinite Library will notify the email address on the acceptance record at least 30 days before a new Sub-processor receives Customer Content. Customer may object in writing within that period on reasonable data-protection grounds; if the parties cannot resolve the objection in good faith, Customer may terminate the affected Service and receive a pro-rata refund of prepaid fees for the unused term.

Infinite Library imposes on each Sub-processor data-protection obligations no less protective than those in this DPA.

8. Personal data breach

Infinite Library will notify Customer without undue delay, and in any event within 72 hours of becoming aware of a personal data breach affecting Customer Content, describing the nature of the breach, the categories and approximate number of data subjects and records concerned, the likely consequences, the measures taken or proposed, and a point of contact. Information may be provided in phases as it becomes available.

9. Audits and information

The Trust Center (Annex IV) is Infinite Library's standing evidence: the sub-processor register, retention schedule, security controls and compliance status are published there and written from the running system. On request Infinite Library will provide its security questionnaire answers and its SOC 2 control map, and will answer Customer's own questionnaire by email at hello@infinitelibrary.ai.

Where the above does not reasonably satisfy an audit obligation under applicable law, Customer (or an independent auditor bound by confidentiality) may audit Infinite Library's compliance with this DPA once in any twelve-month period, on at least thirty days' written notice, during business hours, in a manner that does not disrupt the Services, remotely and document-based first, and at Customer's cost. A supervisory authority's audit rights are unaffected.

10. Return and deletion

Customer's users may export Customer Content at any time from the account dialog (a complete archive of every book, its history and account records). On deletion of an account, Infinite Library locks it immediately and erases Customer Content after a 14-day safety window, save where Union or Member State law requires retention (in which case the data is isolated and retained only for that purpose). Sub-processors delete on their own published schedules (Annex III).

11. International transfers

Infinite Library processes Customer Content in the United States (Annex I). To the extent a transfer of Personal Data from the European Economic Area, the United Kingdom or Switzerland is subject to Chapter V GDPR (or its UK or Swiss equivalents), the parties enter into the SCCs, which are incorporated by reference with Customer as data exporter and Infinite Library as data importer; the optional docking clause applies; Clause 7 is included; the governing law and forum under Clauses 17 and 18 are those of France; Annexes I and II of this DPA serve as the SCC Annexes; and the UK Addendum applies to UK transfers. Where a Sub-processor is certified under the EU-US Data Privacy Framework, Infinite Library may also rely on that certification for the onward transfer.

12. Liability, precedence and term

Each party's liability under this DPA is subject to the limitations and exclusions of liability in the Agreement, except where applicable data-protection law does not permit them to be limited. In the event of conflict, the SCCs prevail over this DPA, and this DPA prevails over the Agreement, in each case for the matters they govern.

This DPA takes effect on acceptance and lasts for as long as Infinite Library processes Customer Content, surviving termination of the Agreement until the last Customer Content is deleted. It is governed by the laws of France.

Annex I — Description of the processing

• Subject matter: the writing, editing, typesetting, illustration, narration, translation, printing and publishing of books from Customer Content, and the storage of those works.
• Duration: the term of the Agreement, plus the deletion window in Section 10.
• Nature and purpose: hosting and processing Customer Content to provide the Services; transmitting it to AI Sub-processors to perform the functions Customer's users invoke; printing and shipping copies on order.
• Categories of data subjects: Customer's personnel and authors; the people described in Customer Content (interviewees, family members, characters based on real people); recipients of printed copies; buyers of Customer's published works.
• Categories of Personal Data: identity and contact details of users and recipients; the content of works, which may contain any category of Personal Data Customer chooses to include; voice recordings volunteered for narration; account and purchase metadata.
• Special categories: none intended. Voice samples for a cloned narration voice are provided voluntarily, can be deleted at any time, and are deleted at the Sub-processor when the voice or the account is deleted.
• Frequency: continuous, for the duration of the Services.
• Location: United States (Railway us-west2 for the studio and workspaces; Google Cloud us-central1 and nam5 for snapshots and accounts), as detailed at the Trust Center.

Annex II — Technical and organisational measures

The measures Infinite Library maintains, grouped as on the Trust Center:

• Encryption: TLS 1.2 or newer on every connection, with HSTS for one year including subdomains.
• Encryption: Encryption at rest for accounts, backups and portraits — Google-managed AES-256 on Cloud Firestore and Cloud Storage.
• Encryption: Card numbers never touch our servers: Stripe holds them (PCI DSS Level 1); we keep only a customer reference.
• Access control: A book is opened only by its owner and the collaborators the owner invites; every collaborator link is a signed, revocable token.
• Access control: Operator actions go through a token-authenticated admin API — short-lived HMAC tokens scoped to one action — and every call is written to an append-only audit log.
• Access control: Secrets live only in the hosting provider's variable store, never in the repository; each token class (print assets, listen links, unsubscribe, sessions, admin) has its own signing key.
• Access control: Production side effects — email, print jobs, purges — are refused on any machine that is not the production container, so a developer laptop with mirrored configuration cannot act on customers.
• Application security: Origin checks on every state-changing request (CSRF) and rate limits on checkout, sign-in, publishing and import routes.
• Application security: Sign-up abuse rails: disposable-domain refusal, velocity caps, and device keys that never store a raw IP address.
• Application security: Account deletion needs a link clicked from the account's own inbox; a stolen session or a social-engineered support chat cannot erase a library.
• Application security: Uploaded files are served under a no-script content-security policy; public pages are server-rendered with no third-party script beyond consent-gated analytics.
• Application security: Dependencies are audited on every release; the last full security review (31 August 2026) found no critical or high-severity dependency issues.
• Resilience: Every change to a book is snapshotted to Google Cloud Storage within seconds; a fresh server re-hydrates from those snapshots on boot.
• Resilience: Health-checked deploys with automatic restarts; a disk-pressure watch reclaims space before a build can corrupt.
• Resilience: Vendor spend ceilings and circuit breakers keep a runaway job from becoming an outage or a bill.
• People and process: A founder-led team: production access is limited to the operator, and the operator's actions are logged.
• People and process: Every change ships through the same typecheck-and-test gate; privacy-critical settings (vendor data-retention flags, the sub-processor register) are pinned by tests that fail when code and policy diverge.
• People and process: Incident response: affected customers — and, where required, supervisory authorities — are notified within 72 hours of confirming a personal-data breach.

Annex III — Sub-processors

The live register, incorporated by reference. Each entry states the vendor, its purpose, the Personal Data it receives, its location, and — for AI vendors — whether Customer Content may be used to train its models and how long it is retained.

• Anthropic (Anthropic, PBC — San Francisco, United States) — The writing and typesetting agent (Claude) — drafting, editing, translation, cover copy, the support concierge, the storybook director. Receives: Your prompts, manuscript, transcripts, notes and uploads, and the book as it is written; the account id as an opaque reference. Location: United States. Training: Never — contractual default. Retention: Inputs and outputs deleted within 30 days (longer only for confirmed policy violations or where the law requires).
• OpenAI (OpenAI, L.L.C. — San Francisco, United States) — Cover art generation, a fallback illustrator, and speech-to-text for voice memos and recordings you upload. Receives: Cover briefs, illustration prompts and reference images, and the audio you ask to transcribe. Location: United States. Training: Never — contractual default. Retention: Abuse-monitoring logs up to 30 days; audio transcription requests are not retained.
• Google Gemini API (Google LLC — Mountain View, United States) — Picture-book illustrations (the storybook engine) and watching YouTube videos you paste as source material. Receives: Illustration briefs (which may name the child a picture book is for), reference images, and the public video links you paste. Location: United States. Training: Never — paid tier. Retention: Logged for a limited period solely to detect abuse (paid tier).
• ElevenLabs (ElevenLabs, Inc. — New York, United States) — Audiobook narration, the optional clone of your own voice, and album music. Receives: The text of chapters you narrate, the voice sample you record for a clone, and song lyrics. Location: United States. Training: Opt-out pending. Retention: Request history is kept by default; zero-retention mode (enable_logging=false) is an enterprise-plan feature.
• Black Forest Labs (Black Forest Labs, Inc. — Freiburg, Germany / United States) — The author photo studio — professional portraits generated from a selfie you choose to upload. Receives: The selfie you upload and the portrait brief. Location: United States / European Union (regional endpoints). Training: Opt-out pending. Retention: No fixed period published — “as long as is reasonably necessary”.
• Railway (Railway Corp. — San Francisco, United States) — Hosts the Bindery studio and the volume that holds book workspaces (manuscripts, sources, uploads, audio masters, chat history). Receives: Everything the studio processes, at rest on its volume and in transit through its edge. Location: United States (us-west2, California).
• Google Cloud / Firebase (Google LLC — Mountain View, United States (Google Ireland Ltd. for EU customers)) — Accounts and sign-in (Firebase Auth), the shared account database (Cloud Firestore), durable workspace snapshots and author portraits (Cloud Storage), and the fallback email transport. Receives: Account profile, plan and credit records, notifications, support conversations, the audit log, book snapshots. Location: United States (Firestore nam5 multi-region; Cloud Storage us-central1, Iowa).
• Stripe (Stripe, Inc. — San Francisco, United States (Stripe Payments Europe Ltd. for EU customers)) — Payments, memberships, refunds, invoices, and author payouts (Stripe Connect). Card numbers never touch our servers. Receives: Email, name, billing country, purchase history, payout details for authors who sell. Location: United States / European Union.
• Twilio SendGrid (Twilio Inc. — San Francisco, United States) — Delivers transactional and lifecycle email (sign-in codes, order updates, release notes). Receives: Email address, name, the content of the messages we send you. Location: United States.
• Lulu Press (Lulu Press, Inc. — Morrisville, North Carolina, United States) — Prints and ships bound copies through its global print network. Receives: Recipient name, shipping address and phone, the print-ready files of the book you order. Location: United States, with printing near the destination.
• Google Analytics 4 & Google Ads (Google LLC — Mountain View, United States) — Product analytics and advertising measurement — asked first in Europe, the UK and Switzerland (Consent Mode v2); cookieless pings until you say yes. Receives: Pseudonymous identifiers and page events; conversions are reported server-side without card or content details. Location: United States.
• ipwho.is / ipapi.co (IP geolocation services (United States / European Union)) — Country lookup for pricing currency and regional rules — the IP address only, nothing else. Receives: IP address. Location: United States / European Union.
• Apple App Store (Apple Inc. — Cupertino, United States) — In-app purchases in the Infinite Library iOS app. Receives: Transaction identifiers (no name or email). Location: United States.
• Vercel (Vercel Inc. — San Francisco, United States) — Hosts the Infinite Library website (www.infinitelibrary.ai), including its guides, reader and the proxied landing. Receives: Request logs; the website's own data as described in the Privacy Policy. Location: United States / global edge.

Annex IV — Trust Center

The Trust Center and its machine-readable twins are part of this DPA's evidence: /trust (this addendum's context, retention schedule, security controls, compliance status), /trust/subprocessors.json (the register), /trust/questionnaire (standing questionnaire answers), /.well-known/security.txt (security contact).

Accept online for your companyAsk a question

Acceptance happens in the studio's account dialog (Privacy & data): sign in, enter the company's legal name and the signatory, and confirm. Your copy — with the fingerprint above — is emailed at once.