Privacy and AI
Claude and privacy: how Cowork handles your business data
Is Claude safe for your business data? With Claude and privacy, what matters most is how you set it up. In Managed Cowork your files and conversations are stored encrypted in our data centre in the European Union, only your question goes to the model, and sensitive work stays with our private model within the EU.
- Your files and chats stay in the European Union
- Sensitive work via the private model in the EU
- Everything traceable in the log
The route of your data
What happens to your data when you use Claude
Privacy with Claude comes down to one question: where does your data go? When you work with Claude in Managed Cowork, your files and conversations stay with us, encrypted on our own hardware in Infodatek Group's Anybase data centre. They do not leave the Netherlands.
For an answer, only your question, with the context the model needs, is sent encrypted to the language model. If you choose Claude, that question goes to Anthropic, under Anthropic's terms. For sensitive data Cowork switches to our private model within the EU, where your data does not go to another party.
That way you use Claude for general work and keep confidential documents in an environment that is truly yours.
Private model
Keep sensitive data within the EU
Not every question needs to go to an online model. Cowork switches between models based on the data you put in. Sensitive data and confidential documents go to our private model, running the DeepSeek or Mistral language model, in a shielded environment that is truly yours, within the EU.
You then use Claude for general work, such as rewriting a text or developing an idea.
- Private model with DeepSeek or Mistral
- A shielded environment that belongs to you
- No other party receives your data
Agreements
Is Claude safe? These agreements apply to every model
The agreements on AI in Cowork apply to Claude just as much as to the other models. They shape how we set up and manage the workspace, and you will find them in the security whitepaper too.
- Your data does not train models
- Answers only come from sources the person asking can access
- Each answer points to its source for checking
- An AI suggestion waits for a person to approve it
- Every question and every answer is logged
Whitepaper
Read it all in the security whitepaper
The whitepaper Private AI in Digital Care explains exactly how Cowork protects your data: where it lives, who can reach it and what the AI does and does not do with it. Two diagrams show the data locations and the route a question takes through the system.
You will find it on the Brochures page, in Dutch, English, French, German and Spanish. You can download it after signing in with your Infodatek account.
- Every conversation in its own isolated workspace
- Application, workspaces and storage in separate network segments
- Connections are only on if you switch them on
- Annual audits for ISO 27001 and NEN 7510
For your own assessment
How to assess Claude for your organisation
- 1
Read the whitepaper
Sign in, download the security whitepaper and review the data locations and the route of a question.
- 2
Walk through it together
We are glad to walk you through the set-up, together with your IT department or security officer if you like.
- 3
Choose models and roles
You decide who works with Claude, who uses the private model and which roles can reach which sources.
- 4
Look back afterwards
The log shows every action and every AI answer, with source, user and time.
This explanation describes how Managed Cowork is set up today and is not legal advice.
Frequently asked questions
Questions about Claude and privacy
Can I use Claude under the GDPR?
That depends on which data you process and how; we do not give legal advice on that. What we do arrange: your files stay in the Netherlands, sensitive work can go to the private model within the EU, and every question is logged. That gives you a solid basis for your own assessment.
Is Claude trained on my business data?
In Cowork, the agreement for every model is that your data is not used to train models. When you work with Claude, your question goes to Anthropic, under Anthropic's terms. If you want certain data to stay within the EU, use the private model for it.
Where are my files and conversations stored?
In the European Union: encrypted, on our own hardware in Infodatek Group's Anybase data centre. Each conversation runs in its own isolated workspace, which shuts down after 30 minutes without activity. Your files are kept.
Can a colleague use Claude to reach data they are not allowed to see?
No. An extra check sits between your systems and the AI, applying the permissions of the person asking. So Claude only answers from sources that person can access.
How do I get the security whitepaper?
On the Brochures page, in five languages. The page is public; you can download after signing in with your Infodatek account, using a code sent by email. The download then starts automatically.
See Digital Care in your own environment
A half-hour conversation with one of our specialists. Your own systems, your own questions. No sales pitch.