Sovereign and EU-hosted
Conversations run on EU-hosted models through our EU inference gateway. Your prompts do not go to a US consumer AI service.
EU-hosted models, no training on your content, and chat history you control. Here is what we do today, and where we stand on certifications.
Sovereign inference in the EU, nothing retained by the models, and only the history you choose to keep.
Conversations run on EU-hosted models through our EU inference gateway. Your prompts do not go to a US consumer AI service.
Our model providers are contractually bound not to store your prompts or outputs, and your content is never used for training.
We keep chat history in your workspace so you can come back to it. You can delete chats at any time. A stricter minimal-retention mode, where new conversations expire automatically, is coming soon.
Read the retention policy ↗The basics, done carefully.
Connections are encrypted in transit.
People and administrators get only the access their role needs.
We log security-relevant events, not the content of your chats.
Report issues to [email protected] and we will investigate.
We design our controls with the requirements of SOC 2 and ISO/IEC 27001 in mind, and we process personal data in line with the GDPR. EUChat.ai does not currently hold a SOC 2 report, an ISO 27001 certificate, or any other third-party certification, and we do not claim HIPAA compliance. If you need evidence for a vendor assessment, ask us and we will tell you exactly what is in place.
Our public policies are online. The DPA and security details are available on request.
How EUChat.ai collects, uses, protects, and gives you control over personal information.
View document ↗POLICY / 02↗The rules for using the EUChat.ai website, workspace, integrations, and AI outputs.
View document ↗AGREEMENT / 03↗Request the DPA for processing personal data on behalf of your organization under applicable data protection law.
Request the DPA ↗ASSURANCE / 04↗Ask for our subprocessor list, technical measures, and current security documentation.
Request security docs ↗Not yet. We align our controls with SOC 2 and ISO/IEC 27001 requirements but hold no certification today. We will say so here when that changes.
No. Your prompts, files and outputs are never used to train or improve any model, ours or a provider’s.
On EU-hosted infrastructure and models. If you connect a third-party tool or model yourself, that provider’s terms apply to that connection.
Yes. Users can delete individual chats or their whole history at any time. See the retention policy.
Yes, on request. Email [email protected].
Tell us what your organization needs to protect and we will answer directly.