This Privacy Policy explains how Anthra (“Anthra”, “we”, “us”) processes personal data when you create an account, write books, narrate chapters, publish them, or otherwise interact with our service at anthra.com (the “Service”). It is written to satisfy the EU General Data Protection Regulation (Regulation (EU) 2016/679, “GDPR”) and the UK GDPR.
1. Data controller
Anthra is the data controller for personal data processed through the Service. Questions about this policy, data-subject requests, and complaints can be sent to will.schulz@aw3.tech. Security disclosures can be sent to will.schulz@aw3.tech.
2. What we collect and why
2.1 Account data
- Name and email address you provide at signup.
- Password hash (Argon2id); we never store your password in clear text.
- OAuth identifiers and tokens when you sign in with Google or GitHub.
- Timestamps recording your acceptance of these Terms and Privacy Policy.
Legal basis: performance of the contract you enter into by creating an account (Art. 6(1)(b) GDPR).
2.2 Content you create
- Book premises, prompts, outlines, prose, illustrations, narration audio, author notes, and style preferences.
- Saved AI-agent conversations and the messages exchanged with the writing assistant.
- Collaborator invites (the email addresses you invite).
Legal basis: performance of the contract; you control whether each book is private, unlisted, or public.
2.3 Operational telemetry
- Sign-in events (event type, OAuth provider, timestamp) so you can see your own activity and we can investigate security incidents.
- Tool-call audit log, generation runs, job records, and token-usage rollups used to debug failures, prevent abuse, and bill credits.
- Rate-limit counters keyed on a short-lived IP-derived identifier (held in memory, not persisted long-term).
Legal basis: our legitimate interest in operating a secure, reliable service (Art. 6(1)(f) GDPR). We retain these records for 90 days and then delete them.
2.4 Reader analytics on published books
When a visitor opens a published book or plays a chapter we log a coarse, pseudonymous identifier derived by HMAC from the visitor’s IP address, user agent, and a daily server-side key that we do not retain. We use this only to count unique readers and identify drop-off points. We do not link this to any individual reader.
Legal basis: legitimate interest (Art. 6(1)(f) GDPR). You can object — see Section 9.
2.5 Cookies
We use cookies that are strictly necessary to keep you signed in and to protect the Service from CSRF attacks. We do not use advertising or cross-site tracking cookies. The product analytics described in Section 2.6 use your browser’s local storage rather than a cookie. See our Cookie Policy for details and controls.
2.6 Product analytics (Viziera)
We use Viziera, a product-analytics service run by AW3 Technology, Inc., which also runs Anthra, and hosted on Vercel and Neon in the USA (us-east-1), to understand how the Service is used and to find bugs and slow pages. Its script loads on anthra.com only after you choose “Accept analytics” in the cookie banner, and then records:
- Page views (the page path, with record ids replaced by a placeholder), sessions, page-load times and Core Web Vitals, and uncaught JavaScript errors.
- A random visitor identifier that Viziera’s script keeps in your browser’s local storage, attached to each of these events.
- When you are signed in: your Anthra user ID (not your email address or name), so that events are linked to your account. We also keep your user ID in local storage (
anthra_viziera_uid) so the link is cleared when you sign out. - From our servers, whatever you chose in the banner (nothing is stored in your browser for these): the amount, currency and purchase reference of credit purchases and refunds, with your user ID; and unhandled server errors (error type, message, stack trace and the route pattern).
As with any website request, Viziera also receives your IP address and browser user agent when your browser sends it data. Viziera keeps ingested data for 400 days. We do not use Viziera for advertising, and we do not sell this data.
Legal basis: for the browser analytics, your consent (Art. 6(1)(a) GDPR), which you can withdraw at any time on our Cookie Policy page; withdrawing deletes the identifiers from your browser. For the server-side purchase and error records, our legitimate interest in understanding and improving the Service (Art. 6(1)(f) GDPR), to which you can object at any time (see Section 9).
2.7 Reports and blocks
Anyone can report a book, chapter, review or author, and signed-in readers can block an author. To run this we store:
- For each report: what was reported, the reason you chose, and the optional note you write (up to 1,000 characters). If you are signed in, the report is linked to your account; if not, it is linked to a keyed hash of a random identifier kept by your browser (in local storage) or by our app, never the identifier itself. Every report also stores a keyed hash of your IP address, never the address itself, so that one person reporting many times from one network counts once. When reports from three different readers are open on a book, it is held off Discover until a moderator reviews it.
- For each block: your account, the blocked author’s account, and the author name and page you blocked from, so we can show you your Blocked authors list. In our apps, a block made while signed out is kept only on your device.
The author of reported content is never told who reported it: if we take something down, the author learns only the category of the report. Authors are not told that you blocked them. Reports are read by Anthra’s moderators, and the report and your note are emailed to our support inbox.
Legal basis: our legitimate interest in keeping the public library safe and meeting app-store rules for user-generated content (Art. 6(1)(f) GDPR); for blocks, performance of the contract.
3. How long we keep data
- Account data: while your account is active, plus up to 30 days after deletion to allow reversal of a destructive request.
- Books and chapters: until you delete them, with a 30-day soft-delete window before permanent removal.
- Agent conversations: until you delete them or your account.
- Operational telemetry: 90 days, then permanently deleted by a daily cron job.
- Reader analytics: 90 days for individual events, aggregated indefinitely.
- Product analytics (Viziera): up to 400 days, as stated by Viziera.
- Reports: there is no fixed deletion schedule; a report is kept as moderation history after it is resolved. It is deleted when the reported book or its author’s account is deleted. If you delete your account, reports you filed are kept, but the link to your account is removed.
- Blocks: until you unblock the author, or either account is deleted.
- Backups: rolling encrypted backups for up to 30 days.
4. Who we share data with (sub-processors)
We use carefully selected service providers (“sub-processors”) under Article 28 GDPR data-processing agreements. The current list:
- Vercel Inc. (USA) — hosting, edge network, function execution, blob storage for cover images and audio, and the AI Gateway route used for image upscaling.
- Neon (Databricks Inc.) (USA/EU) — managed PostgreSQL database.
- Anthropic, PBC (USA) — Claude language models for outlines, prose, editing tools and the copilot.
- OpenAI, L.L.C. (USA) — image generation for covers and illustrations.
- ByteDance — the Seedream model for image upscaling, reached through the Vercel AI Gateway.
- Perplexity AI Inc. (USA) — web research using the premise and genre of a new book.
- ElevenLabs, Inc. (USA) — text-to-speech narration.
- Resend, Inc. (USA / EU region) — transactional email (verification, password reset).
- cred.diy (USA, on Supabase) — credit ledger and checkout for paid usage.
- Lulu Press, Inc. (USA) — print-on-demand orders, including your name and shipping address when you order a printed copy.
- Apple Inc. (USA) — push notifications to our apps, if you allow them.
- GitHub, Inc. and Google LLC (USA) — optional OAuth identity providers.
- Viziera (viziera.ai; run by AW3 Technology, Inc., which also runs Anthra; hosted on Vercel and Neon, USA) — product analytics: page views, sessions, performance and error reports from the website if you accept analytics, purchase and refund amounts, and server errors, linked to your user ID when you are signed in (Section 2.6).
A current, dated sub-processor list is maintained in our public repository and is available on request. We notify you at least 14 days before adding a new sub-processor that handles personal data.
5. International transfers
Most of our sub-processors are based in the United States. Where personal data is transferred outside the EEA / UK we rely on the Standard Contractual Clauses (SCCs) approved by the European Commission and, where applicable, the EU–US Data Privacy Framework. A copy of the relevant SCCs and a description of supplementary measures can be obtained from will.schulz@aw3.tech.
6. Security
We protect personal data with industry-standard controls: TLS in transit; encrypted storage at the sub-processor layer; Argon2id password hashing; per-tenant access scoping; rate limiting; CSRF defenses; isolated secret storage; and least-privilege access for engineers. We disclose confirmed personal-data breaches to the competent supervisory authority within 72 hours where required and notify affected users without undue delay.
7. AI provider data handling
Prompts, premises, and chapter prose are sent to Anthropic, OpenAI, Perplexity, and ElevenLabs to produce the requested output. We call these providers directly. Images you upscale go through the Vercel AI Gateway to ByteDance’s Seedream model.
Anthropic’s commercial terms do not allow it to train on this content, and it deletes API inputs and outputs within 30 days, with exceptions for usage-policy enforcement and legal requirements. OpenAI does not train on API data unless a customer opts in, and keeps abuse-monitoring logs for up to 30 days. Perplexity’s API terms do not allow it to train on this content. ElevenLabs keeps generation history by default. We have opted our ElevenLabs workspace out of model training, which applies from November 6, 2026 and does not affect earlier use.
8. Children
Anthra is not directed to children under 16, and we do not knowingly collect personal data from anyone under that age. If you become aware that a child has provided us with personal data, contact will.schulz@aw3.tech and we will delete it.
9. Your rights
Under the GDPR and UK GDPR you have the right to:
- Access the personal data we hold about you — available from your account page as a JSON export. It includes the authors you blocked and the reports you filed (with your notes). It does not include reports other people filed about your work, since they would identify the reporter.
- Rectify inaccurate data — change your name, email, and content directly from your account.
- Erase your data — delete your account from your account page. We hard-delete account, books, chapters, agent conversations, audit records, operational telemetry, your blocks, and reports about your work within 30 days and instruct sub-processors to do the same. Reports you filed are kept as moderation history with the link to your account removed.
- Restrict or object to processing based on legitimate interest — email will.schulz@aw3.tech and we will action the request within 30 days.
- Withdraw consent at any time without affecting the lawfulness of processing before withdrawal.
- Data portability — the JSON export above is in a structured, commonly used, machine-readable format.
- Lodge a complaint with your local supervisory authority. EU residents can find theirs at edpb.europa.eu; UK residents can contact the Information Commissioner’s Office at ico.org.uk.
10. Marketing email
We send transactional email (verification, password reset, billing receipts) on the legal basis of contract performance and cannot opt out of these while you have an account. We send marketing email only if you explicitly opted in at signup; you can unsubscribe at any time using the link in any marketing email.
11. Changes to this policy
If we make material changes we will notify signed-in users by email and in-product, at least 14 days before the change takes effect. Material changes will require renewed acceptance. The effective date at the top of this page is the date of the current version.