Dieses Dokument liegt nur auf Englisch vor.
Dokumente
Sub-processors
The Chattopus service, provided by ERBA INTERACTIVE. Version 1, last updated on 19 November 2026.
What this page is for
When you use Chattopus, ERBA INTERACTIVE processes on your behalf the data of the conversations in your website's chat, as a processor (Article 28 GDPR). For some activities we use other providers, the sub-processors. This page lists all of them, with what they do, what data they receive and where they process it.
This page is Annex 3 to the Data Processing Agreement (DPA) and makes available the identity of everyone in the chain, as recommended by the European Data Protection Board (Opinion 22/2024).
How we tell you about changes
- At least 30 days before a new sub-processor starts processing your data, we send an email to the users with the Owner role in your organisation and we update this page, including the change log at the bottom.
- Within 15 days of the notice you can object on reasonable grounds related to data protection, by writing to assistance@chattopus.com. If we do not find a solution, you can withdraw from the contract without penalty and we refund the part of the fee already paid and not yet used (DPA, Article 8.3).
- Only in an emergency (for example the sudden closure of a service we depend on) may the notice be shorter: in that case we tell you as soon as possible, and your right to object stays the same.
- If you want to receive the notices at an address other than the Owners' addresses, write to assistance@chattopus.com.
1. Sub-processors that process conversation data
| Sub-processor | Activity | Data processed | Place of processing | Safeguards for transfers outside the EEA |
|---|---|---|---|---|
| OVHcloud | Application server and database of the Service | All the data of the Service: conversations (with the visitor data that comes with them, such as the IP address for 30 days, and the team notes), contacts, knowledge base, settings, Operators' accounts, customers' artificial intelligence keys and the secret keys for their websites' registered customers (encrypted) | Gravelines, France (European Union) | Not needed: the data stays in the EEA |
| Anthropic | The dashboard's Help, the demo and the chat on our public website (for that chat we are the controller): Help is an automatic assistant that receives no data of the visitors of customers' websites (see below) | For Help: the questions of dashboard users, essential information about the account and a summary of how the websites are set up (no visitor data); for the chat on our website: the messages of the people who write to us there; never data of the visitors of customers' websites | Outside the EEA: the United States or worldwide (the “us” or “global” processing options of Anthropic's API) | The EU-US Data Privacy Framework or standard contractual clauses |
| Brevo | Sending the service emails: the emails to the visitor with the team's reply; notices about payments, the yearly renewal, the cancellation and the reactivation of a subscription, a plan not yet chosen, an artificial intelligence key that does not work and the deletion of data; the codes to sign in, to create an account and to join a team, the notice that an account with that email already exists, invitations, password resets and the link to use the dashboard on a phone; support requests sent from the dashboard's Help or from our website's contact form, and our support's replies to them. Conversations are never emailed to Operators | Email addresses of Operators, for the emails about their account; in support requests and our replies, the name and email address of whoever wrote and their request; in the emails to the visitor, the visitor's email address and name, if they left it, the name of the business (also as the sender's name) and its address for the visitor's replies (“Email for your customers’ replies”), the name of the team member who replied, the team's latest replies (three at most) and the link back to the chat. Each email has a single recipient | As set out in Brevo's data processing agreement | As set out in Brevo's data processing agreement |
The artificial intelligence provider you choose
The chat's artificial intelligence is the one you choose: you create an account with one of the providers the dashboard lets you connect (Anthropic for Claude, OpenAI for ChatGPT, Google for Gemini, xAI for Grok, Mistral AI, DeepSeek, Groq, OpenRouter or another service compatible with the OpenAI APIs) and you connect its key in the dashboard, website by website. That provider is not one of our sub-processors and does not appear in this list:
- you choose it and the contract is between you and the provider: you decide the provider, the model, the settings of your account (for example data retention and use for training) and where it processes the data;
- we send it the data needed for each reply (system instructions, extracts from the knowledge base, the conversation history, the path and title of the page; never the IP address or the session token, nor the other data of the visitor card or the team notes) only because you ask us to, by connecting the key (DPA, Article 8.9);
- name it in your privacy notice for visitors: our notice template has room for it;
- if you remove the key from a website (and its Backup AI, if any) or switch off that website's artificial intelligence, we no longer send anything for that website: its chat answers with the automatic replies from your FAQs and with your team.
Notes on the platform's artificial intelligence provider
For the cases in the table we use Anthropic's API, called directly from our servers.
- Contracting entity and region. According to our research of 25 September 2026, the direct API processes data only with the “global” or “us” options: this is a transfer outside the EEA, to be covered by the Data Privacy Framework or by standard contractual clauses.
- Alternatives with processing in the EU, if we choose them in the future: Amazon Bedrock (for example the eu-south-1 region, Milan, with routing restricted to the EU: according to our research, requests stay in EU countries, but there is no processing in Italy only) or Google Cloud Vertex AI with an EU endpoint; both providers belong to US groups. In that case we will update this page with the advance notice described above.
For the artificial intelligence you choose, your provider's terms apply.
What we do not use
- The search in the knowledge base and the matching against the FAQs for automatic replies take place on our servers with a text index: we do not use external services for text analysis, translation or semantic indexing.
- The country a visitor writes from is estimated on our servers from their browser's time zone, and the device from the information the browser sends (the user agent): we do not use geolocation databases or external services, and no IP address is sent to anyone else for this.
- The chat does not set cookies and does not load third-party analytics or advertising tools.
2. Services that do not receive conversation data
They are not sub-processors for visitor data, but we list them for transparency.
| Service | Why it is listed here | Data involved |
|---|---|---|
| Browser push notification services (Apple, Google, Mozilla, Microsoft) | They deliver push notifications to Operators. The service used depends on the Operator's browser | A message encrypted under the Web Push standard, with a title, a generic text, technical identifiers and a link to the dashboard. No visitor data |
| Paddle (Paddle.com Market Ltd, United Kingdom) | Sells the subscriptions to the customer as a reseller (merchant of record): collects the payments, adds VAT, sends receipts and invoices, handles refunds and disputes | Data of the buyer (the customer's name, email address, postal address and payment details), not of visitors. Paddle processes it as the seller, probably as an independent controller |
| The artificial intelligence provider chosen by the customer | Receives the data needed for each reply on the customer's instructions, under its own contract with the customer (section 1) | The data listed in section 1, “The artificial intelligence provider you choose”, for the artificial intelligence's replies |
3. For agencies: the whole chain
If the website is managed by a web agency that uses Chattopus under its own brand, the chain is:
End Client (controller)
└─ Agency (processor)
├─ ERBA INTERACTIVE (sub-processor)
│ └─ the sub-processors listed in section 1 (further sub-processors)
└─ the artificial intelligence provider chosen by the Agency (chosen and contracted by the Agency)
The Agency must make this list available to its End Clients, forward every notice of a change to them and name in its own list the artificial intelligence provider it has connected (Agency and Reseller Addendum, Article 5).
4. Change log
| Date | Change |
|---|---|
| 19 November 2026 | First publication of the list (version 1). |
Questions about this page: assistance@chattopus.com. Permanent address: chattopus.com/sub-processors.