A team uses an AI chat in line with data protection when it runs through a company account with a contract instead of private accounts and everyone knows which data may go in. If the provider processes personal data on your behalf, you need a data processing agreement under Art. 28 GDPR. Part of it is knowing where the data lives and which providers are involved.
Why are private AI accounts a problem?
Many people have long used AI at work with their own tools. According to the Work Trend Index by Microsoft and LinkedIn, 75 percent of knowledge workers worldwide use generative AI at work, and 78 percent of these users bring their own AI tools.
A private account sits outside your administration and has no contract with your company. Whatever quotes, customer data or prices end up there, you neither see nor control.
Source: Microsoft and LinkedIn: 2024 Work Trend Index Annual Report, 8 May 2024
What does a data processing agreement cover?
If a provider processes personal data on your behalf, Art. 28 GDPR requires a contract. Among other things it covers the subject, duration, nature and purpose of processing, the types of data, your instructions, confidentiality, security measures, sub-processors and deletion at the end.
Ask every provider for this contract and for a list of the sub-processors it uses before your team starts.
Source: European Union, EUR-Lex: Regulation (EU) 2016/679, General Data Protection Regulation, 27 Apr 2016
Where is the data processed?
Ask where the data is stored and in which data centres the AI models run. Transfers to countries outside the EU need a basis of their own under the GDPR.
With Agentiq Chat: your data lives on servers in Germany, and we use AI only in data centres in the EU. We do not add models without EU processing, and your data is not used for training.
Source: European Union, EUR-Lex: Regulation (EU) 2016/679, General Data Protection Regulation, 27 Apr 2016
Which rules does the team need?
Decide which data may go into the chat: personal data only when the task requires it, health data and other specially protected data only with a clear legal basis.
And decide that a person checks what the AI writes. According to Bitkom, 42 percent of those who use AI for search have already received false or invented information.
Source: Bitkom: Internet search in transition: half already use AI chats, 20 Nov 2025
May admins read the chats?
Settle that beforehand and tell your team. In Agentiq Chat admins manage access and settings; they do not see individual chats.
If you have a works council, involve it early.
How does Agentiq Chat help?
One AI chat for the whole company, processed in the EU.
- All major models, processed in the EU
- Projects with your own knowledge and documents
- No training on your data
Price on request.
Frequently asked questions
Is ChatGPT forbidden at work?
No. What matters is how it is used: with which account, under which contract and with which data.
Is our data used for training in Agentiq Chat?
No. Your chats, projects and documents are not used to train models.
Where do I get the data processing agreement?
With your booked product you receive the agreement with the fitting annexes. The list of providers on our website names every provider used with purpose and region.
Do I need a data protection impact assessment?
That depends on which data you process and to what extent. Settle it with your data protection officer. This text is not legal advice.
What does Agentiq Chat cost?
Agentiq Chat comes for up to 5, 25 or 100 people. We give the prices for Small, Medium and Premium on request. The purchase page always shows the current offer.