← Back to Deep Research CIATA GmbH

Privacy Policy

Deep Research — AI-Powered Research Reports

Last updated: July 15, 2026 · CIATA GmbH, Berlin

This Privacy Policy applies to CIATA GmbH, the Deep Research web application at research.ciata.io, and all related services (together the "Services"). CIATA GmbH acts as the controller within the meaning of Art. 4(7) GDPR.

1. Controller

CIATA GmbH
Arnimallee 7
14195 Berlin, Germany
HRB 266084 B · Amtsgericht Charlottenburg
Managing Director: Prof. Dr. Tim Landgraf
Email: privacy@ciata.io

2. Principles of Processing

We process personal data in accordance with the principles of lawfulness, fairness, transparency, purpose limitation, data minimisation, accuracy, storage limitation, and integrity and confidentiality pursuant to Art. 5 GDPR.

All personal data is processed exclusively within the European Economic Area (EEA). For website statistics we use only a self-hosted, cookie-free analytics tool (Plausible); we do not use any third-party marketing, advertising, or tracking services. No personal data is transferred to countries outside the EEA.

3. Categories of Personal Data

We process the following categories of personal data:

  • Account data — email address, hashed password
  • Usage data — research questions submitted, reports generated
  • Technical data — IP address, browser type, timestamps
  • Payment data — transaction records (processed via Mollie; we do not store card details)

4. Legal Bases for Processing

Each processing activity is based on one of the following legal bases:

  • Art. 6(1)(b) GDPR — performance of a contract (account creation, research execution)
  • Art. 6(1)(f) GDPR — legitimate interests (IT security, error resolution, platform stability)
  • Art. 6(1)(c) GDPR — compliance with legal obligations (tax, accounting)

We do not process personal data on the basis of consent for marketing or advertising purposes. We do not use cookies beyond what is technically necessary for session management.

5. How the Service Works — Data Flow

When you submit a research question, the following processing takes place. We describe this in detail so you can understand exactly where your data goes.

✓ EU AI Processing

Your research question and all AI inference run on AKI.io, a German-hosted LLM provider (Frankfurt, Germany). No data leaves EU servers for AI processing. Your queries are never used for model training.

Provider
AKI.io GmbH — Frankfurt, Germany 🇩🇪
Data sent
Research question, page excerpts for analysis
Training
No — user data is never used for training
Legal basis
Art. 6(1)(b) GDPR (contract performance)

✓ EU Web Search (Default: Staan)

By default, derived search queries are sent to Staan, an EU-sovereign search index operated by Ecosia and Qwant. All query data stays within EU jurisdiction. The search engine sees only derived queries — never your original question.

Provider
European Search Perspective GmbH (Ecosia + Qwant) — EU 🇪🇺
Data sent
Derived search queries (not your original question)
Legal basis
Art. 6(1)(b) GDPR (contract performance)

✓ EU Page Fetching

Web pages are fetched by EU-based servers, not by your browser — either directly by our server in Germany or, in the default search mode, by our EU search provider (Staan), which retrieves the page content and relays it to us. Third-party websites only ever see an EU server’s IP address and standard HTTP headers. Your personal IP, identity, and location are never exposed to third-party websites.

Fetched by
Our server (Hetzner, Germany 🇩🇪) or, by default, Staan (EU 🇪🇺) — never your browser
Data exposed to third parties
An EU server’s IP + HTTP headers only

✓ EU Data Storage

All research results, reports, user accounts, and intermediate data are stored locally on the server in Germany. No cloud storage services are used.

Storage
SQLite database on the server
Location
Hetzner Cloud — Germany 🇩🇪
Encryption
Passwords hashed with bcrypt; all connections encrypted in transit (TLS). Backups are encrypted and stored geo-redundantly within the EU (Frankfurt → Helsinki).

6. Recipients and Categories of Recipients

We disclose personal data only where necessary for the purposes described in this Privacy Policy:

  • AKI.io GmbH — AI inference provider (processor; data processing agreement pending execution)
  • European Search Perspective GmbH (Staan) — search provider (derived queries only)
  • Hetzner Online GmbH — infrastructure / hosting provider
  • Mollie B.V. — payment processing (Amsterdam, Netherlands 🇳🇱)
  • Billie GmbH — Berlin, Germany 🇩🇪. Billie is a B2B "pay by invoice" payment method offered within Mollie's checkout (we have no direct integration with Billie); if you choose it, Billie processes your billing data as an independent controller for its own credit/underwriting decision
  • Qonto — invoice generation (Paris, France 🇫🇷); receives your name, email and billing address to issue an invoice for each purchase
  • Brevo (Sendinblue) — transactional email delivery (Paris, France 🇫🇷)
  • Tax advisers and accountants as required by law

Processors that act on our behalf (e.g. AKI.io, Hetzner) do so only on our documented instructions. The Art. 28 GDPR data processing agreements with these processors are currently being finalised and are not yet executed; until they are, the instruction-only basis described here is contractual intent rather than a concluded agreement. Our payment partners — Mollie, and, for the business "pay by invoice" option, Billie and Qonto — act as independent controllers under their own regulatory obligations; we disclose them here and share only the data necessary for payment and invoicing.

Our service enforces strict per-user access controls: no account — including administrator accounts — can view another user's research questions or reports through the application. At the infrastructure level, a limited number of authorised technical staff may access stored data only where strictly necessary for hosting operation, security, backups, and the resolution of technical faults or legal obligations. Such access is limited to the minimum necessary and subject to confidentiality obligations. In any event, research questions and reports are automatically deleted after 30 days (see Section 14), after which no such data remains available to us. Dominant legal basis: Art. 6(1)(f) GDPR (legitimate interest in operating and securing the service).

7. Account Registration

For registration and authentication, we process your email address and a hashed password. Passwords are hashed with bcrypt and never stored in plaintext.

Dominant legal basis: Art. 6(1)(b) GDPR.

Providing your email address is not a statutory requirement, but it is necessary for the conclusion and performance of the contract: without it we cannot create an account or provide the service. Providing billing details is likewise required to complete a purchase, and — where an invoice is issued — to meet our statutory record-keeping obligations (Art. 13(2)(e) GDPR).

8. Billing and Payments

Payment processing is handled by Mollie B.V., an EU-native payment provider headquartered in Amsterdam, Netherlands. Mollie is regulated by De Nederlandsche Bank (DNB) and processes all payment data within the EU. We do not store credit card numbers or bank details on our servers.

Dominant legal basis: Art. 6(1)(b) GDPR; for statutory retention, Art. 6(1)(c) GDPR.

9. Transactional Emails

We use Brevo (formerly Sendinblue), headquartered in Paris, France, to send transactional emails such as account verification, password resets, and payment confirmations. For this purpose, your email address is shared with Brevo. Brevo processes data exclusively within the EU and is subject to a data processing agreement.

Dominant legal basis: Art. 6(1)(b) GDPR.

10. Technical Logs and Security

Technical logs (IP addresses, timestamps, request metadata) are processed solely to ensure platform security, stability, and error resolution. Logs are automatically deleted after 90 days.

Dominant legal basis: Art. 6(1)(f) GDPR.

11. Privacy-Friendly Analytics — No Advertising, No Tracking Cookies

To understand aggregate traffic, we use Plausible, a privacy-friendly analytics tool that we self-host on our own EU infrastructure (trc.ciata.io). Plausible sets no cookies, collects no personal data, and shares nothing with third parties.

We do not use Google Analytics, Matomo, advertising services, tracking pixels, or third-party cookies. The only data stored in your browser is a session token (JWT) for authentication, held in localStorage.

12. AI Model Training

Your research questions, results, and reports are never used to train AI models — neither by us nor by our AI provider (AKI.io). AKI.io processes prompts exclusively in volatile memory and does not log, store, or analyse customer-provided content.

13. AI Act Transparency Measures

In compliance with EU AI Act Article 50(2), all research reports generated by Deep Research are marked as AI-generated content through three complementary mechanisms:

  • Visible labelling — every report and PDF export displays a clear "AI Generated" indicator.
  • Invisible digital watermark — each report carries a machine-readable watermark using the VSRMark steganographic algorithm (Unicode Variation Selectors). This watermark is invisible to human readers but can be detected by software.
  • Machine-readable provenance block — every report begins with an AI-PROVENANCE comment recording the generator, our organisation, an ai_generated flag, the AI models that processed the request, the generation timestamp, and a truncated hash of the research question.

The embedded watermark contains only technical metadata:

  • Producer identifier ("CIATA Deep Research")
  • A watermark format version (a fixed protocol identifier, not the software release number)
  • An ai_generated: true flag
  • The internal job identifier (a random alphanumeric string)
  • The UTC timestamp of report generation

No personal data (name, email, IP address, research question) is embedded in the watermark. The watermark cannot be used to identify the user who commissioned the report.

The provenance block additionally contains a question_hash — the first 16 hexadecimal characters of a SHA-256 hash of your research question. It is a one-way value used to tell two reports apart; the question cannot be reconstructed from it, and it does not identify you.

Dominant legal basis: Art. 6(1)(c) GDPR (legal obligation under EU AI Act).

14. Data Retention

Personal data is deleted or anonymised once the processing purpose no longer applies:

  • Account data — retained until account deletion
  • Research reports — automatically deleted after 30 days, or earlier if you delete them or request erasure
  • Technical logs — automatically deleted after 90 days
  • Payment records — retained for 10 years per statutory tax obligations (§ 147 AO)

15. Your Rights

You have the following rights under the GDPR:

  • Right of access (Art. 15 GDPR)
  • Right to rectification (Art. 16 GDPR)
  • Right to erasure (Art. 17 GDPR)
  • Right to restriction of processing (Art. 18 GDPR)
  • Right to data portability (Art. 20 GDPR)
  • Right to object to processing based on Art. 6(1)(f) GDPR (Art. 21 GDPR)

To exercise any of these rights, contact us at privacy@ciata.io.

16. Supervisory Authority

You have the right under Art. 77 GDPR to lodge a complaint with a data protection supervisory authority — in particular in the Member State of your habitual residence, your place of work, or the place of the alleged infringement. The authority responsible for us is:

Berliner Beauftragte für Datenschutz und Informationsfreiheit
Friedrichstraße 219
10969 Berlin, Germany
Phone: +49 30 13889-0
Email: mailbox@datenschutz-berlin.de
www.datenschutz-berlin.de

17. Automated Decision-Making

No automated decision-making within the meaning of Art. 22 GDPR takes place. AI-generated research reports are informational outputs, not decisions that produce legal or similarly significant effects.

18. Amendments

This Privacy Policy may be updated from time to time. We will notify registered users of material changes by email. Changes are effective when posted on this page.