Mistral AI is French, its servers sit in the EU and the company was founded in Paris - so GDPR compliance should be settled, right? Not quite. The answer depends on whether your team uses the free Le Chat, the API with a data processing agreement or just pastes customer data into a browser window without telling anyone. Below is the plain picture, as of November 2026: what Mistral offers today, where the gaps are, what it costs, and how a ready-made AI team from Superkind puts Mistral-class models to work in your programs across 1,700+ integrations - hosted in Europe. Not legal advice - confirm the details with your data protection officer.
Key takeaways
Get a ready-made AI team for everyday work - Superkind comes with Inbox Sorter, Contract Watcher and Ticket Resolver that triage your inbox, track contract deadlines and answer support tickets. Start for free with unlimited users, or build your own.
Keep the control, keep the data in Europe: hosted and processed in Europe, GDPR compliant, aligned with the EU AI Act (Regulation (EU) 2024/1689), no training on your data, DPA via the Trust Center, and you decide which steps wait for your OK68.
Mistral is European, but the plan decides: the API and Enterprise plans come with a DPA, EU hosting and no training on your data by default. Le Chat free runs under consumer terms without a business DPA, and conversations may be used for model improvement12.
Your duties stay with you: a lawful basis (Art. 6 GDPR), a record of processing (Art. 30), a DPIA where the risk is high (Art. 35) and AI literacy under Art. 4 of the EU AI Act56.
Your AI team for everyday work
Is Mistral GDPR-compliant?
Mistral AI is a French AI company (Mistral AI SAS, Paris) that builds large language models and offers them through Le Chat (a consumer chat interface), a developer API and enterprise agreements. Because Mistral is headquartered in the EU and processes data on European infrastructure, it starts in a stronger position than US or Chinese providers on the GDPR transfer question17. That does not make every use automatically compliant: Le Chat free has no business DPA, conversations may be used for training unless the user opts out, and your company still needs a lawful basis, a record of processing and AI literacy under Art. 4 of the EU AI Act256. An example of an AI employee that wraps Mistral-class models in a team layer with approval and company knowledge is Superkind, which works in Outlook, Microsoft Teams, Slack and the web app through 1,700+ integrations including Pipedrive, Notion and Zendesk8.
European does not mean automatically compliant
Mistral’s EU base means no Art. 44 third-country transfer for the data itself - a real advantage. But GDPR compliance is more than a server address: you still need the DPA (Art. 28), training opt-out, the record of processing (Art. 30) and staff rules under the EU AI Act156.
How it works with Superkind: your AI team for everyday work
You write to Superkind by email, in Microsoft Teams, Slack or the web app, like a colleague, and Superkind does the task in your programs through 1,700+ integrations, among them Pipedrive, Notion and SharePoint. It works with your tools instead of replacing them, hosted in Europe and GDPR compliant.
With Superkind you don’t start from zero: it comes with ready-made AI employees for the work a general chatbot cannot finish. They run on their own in the background, and you decide where they wait for your OK. Use them right away, adapt them, or build your own.
Inbox Sorter
AI employee for email triage in Outlook
The shared inbox gets forty new messages overnight and no one knows which five are urgent. Inbox Sorter reads every email, matches it to the right person and project in Pipedrive, labels it, drafts a reply where one is obvious and flags the rest. By the time the team opens Outlook the queue is sorted.
Contract Watcher
AI employee for contract deadline tracking
Three supplier contracts auto-renewed last quarter because nobody saw the notice period. Contract Watcher reads every new contract in SharePoint, pulls the notice period and renewal date, logs them in Notion and posts a Slack reminder thirty days before each deadline. No missed window again.
Ticket Resolver
AI employee for support ticket triage in Zendesk
Half the support tickets are the same five questions: login reset, invoice copy, delivery status. Ticket Resolver reads the ticket in Zendesk, searches your Notion help centre and Company Brain, drafts the answer and waits for your OK on anything that touches billing. The team handles only what actually needs a person.
Every employee is ready to use right away. Or build your own on the platform.
Safe by default
Mistral is European, but a raw API still needs the team layer: approval, company knowledge and a DPA. With Superkind you get all three out of the box.
- Hosted and processed in Europe
- GDPR compliant
- Aligned with the EU AI Act
- No training on your data
- DPA through the Trust Center
What Mistral offers today
- Le Chat (free and Pro) - a consumer chat interface for conversations, web search and document uploads. Conversations may be used for model training unless the user opts out in the settings2.
- API (La Plateforme) - pay-per-token access to Mistral Large, Mistral Small, Codestral and other models. Mistral offers a DPA for API customers and states that customer inputs and outputs are not used for training by default17.
- Enterprise agreements - custom plans with SLA, dedicated support, volume pricing and a signed DPA; the same no-training-by-default policy as the API1.
- EU hosting by default - Mistral processes inference in Europe, which removes the Art. 44 third-country transfer that US providers must address with Standard Contractual Clauses17.
- Open-weight models - Mistral 7B, Mixtral and other models are published under permissive licences and can be self-hosted on your own infrastructure3.
What is missing for company use
Mistral covers the hosting and the DPA on commercial plans. The gaps are in practice.
- Le Chat free has no business DPA - consumer terms, no central admin, training opt-out per user. Customer data in a free Le Chat account sits outside your compliance setup2.
- Training opt-out is per user on Le Chat - if someone in your team forgets to switch it off, their conversations feed model improvement2.
- An API alone does not do the work - Mistral gives you a model endpoint. It does not read the account in Pipedrive, write the reply in Outlook, log the thread in Notion or wait for a human OK before sending. You build that yourself or use a tool that does both.
- Your own paperwork - a lawful basis under Art. 6 (and Art. 9 for special categories), an entry in the record of processing (Art. 30), a DPIA (Art. 35) where the risk is high, and AI literacy training under Art. 4 EU AI Act56.
- No approval layer - the API returns text. Whether that text goes straight to a customer or waits for review is up to your integration. Approval is not built in.
Mistral and alternatives compared
| Option | Hosting / Data protection | Training on your data | DPA / AVV |
|---|---|---|---|
| Superkind | EU hosted; 1,700+ integrations incl. Pipedrive, Zendesk, Notion | Never | Yes, via the Trust Center |
| Mistral Le Chat free | EU hosted (Paris)7 | Yes, unless user opts out2 | No business DPA2 |
| Mistral API | EU hosted17 | No by default1 | Yes1 |
| Mistral Enterprise | EU hosted, dedicated support1 | No by default1 | Yes, signed1 |
As of November 2026. Sources: Mistral terms1, privacy policy2, trust page7, Superkind Trust Center8. See also Is ChatGPT GDPR-compliant? and Is Claude GDPR-compliant?
Examples: Mistral and the GDPR in practice
1. Superkind: European models at work in your programs
Daniel writes in Teams: “Answer the pricing question from Solaris SRL and log it in Pipedrive.” Inbox Sorter reads the email, pulls the account and the last offer from Pipedrive, drafts the reply and waits for Daniel’s OK. Contract Watcher flags a supplier deadline in Slack. Ticket Resolver closes a tier-1 ticket in Zendesk. All processed in Europe with models from OpenAI, Anthropic, Mistral and Google, through 1,700+ integrations8. See also email inbox triage.
2. A developer builds on the Mistral API with a DPA
The team connects Mistral Large through the API, signs the DPA, confirms no training on inputs by default and documents the setup in the record of processing under Art. 3015. What is left: building the integration to your systems, the approval flow and the staff rules yourself.
3. A colleague uses Le Chat free for customer emails
Someone pastes a customer complaint into a free Le Chat window. That conversation may be used for model improvement, there is no business DPA and the company has no oversight2. The fix: make a rule for staff that company data only goes through company accounts - the API with a DPA, or a tool like Superkind.
4. The marketing team uses Mistral for campaign copy
Marketing drafts social-media posts in Le Chat Pro. No personal data is in the prompts, the risk is low, and an opt-out from training is switched on. This is a low-risk creative use. The team still adds Mistral to the record of processing and trains staff under Art. 4 EU AI Act56.
What a compliant setup needs (and what it costs)
- Use commercial accounts for company data - the API or Enterprise plan with a DPA. Le Chat free is not the right place for customer or employee data12.
- Switch off training where it matters - on Le Chat, opt out in the settings. On the API, training is off by default; confirm it in the DPA2.
- Record of processing and DPIA - add Mistral to your Art. 30 record, run a DPIA under Art. 35 where the risk is high5.
- AI literacy and transparency - train staff under Art. 4 EU AI Act; Art. 50 adds transparency duties for some AI-generated content from August 20266.
- An approval layer for real work - the API returns text, it does not ask before sending. Add approval yourself or use a tool that has it built in.
What Superkind costs
As of November 2026. Prices excl. VAT, for the whole company with unlimited users.
- Free - EUR 0 with a one-off up to 150 hours of work free to try. No credit card, no sales call, every feature and every connection, in Slack, Teams, email and the app, hosted in the EU.
- Business - from EUR 50 per month for 50 hours of work a month. EUR 100 is the popular step, then EUR 200 and EUR 500. No price per seat, unused hours carry over one month, custom apps included, cancel monthly.
- Enterprise - custom price for your volume, dedicated support with SLA, SSO/SCIM and audit log, a DPA and help with the security review, payment by invoice.
How to start
Your 3-step plan
- Write the rule for staff - company data only in company accounts with a DPA. Put it in writing and cover Art. 4 EU AI Act AI literacy with a short training56.
- Give Superkind one real task - sign up for free at app.superkind.ai with up to 150 hours of work, send one real task in Outlook, Teams or Slack and see Inbox Sorter, Contract Watcher or Ticket Resolver do it, all processed in Europe8.
- If you build on the Mistral API directly - sign the DPA, confirm no training, add it to your record of processing and build the approval flow and system connections yourself15.
Conclusion
Mistral AI has the strongest starting position of any large-model provider for GDPR: French headquarters, EU hosting, DPA on commercial plans and no training on API data by default17. The gap is not in the model but in what sits around it: Le Chat free has no business DPA, an API alone does not do the follow-up work in your programs, and the company’s own documentation duties stay with you. If you want the work done from the first task - emails triaged, contracts watched, tickets resolved - Superkind brings a ready-made AI team (Inbox Sorter, Contract Watcher, Ticket Resolver) that runs through 1,700+ integrations including Pipedrive, Zendesk and Notion, with approval where you want it and all processing in Europe. Write the staff rule and send Superkind one real task this week.
Frequently Asked Questions
Mistral AI is a French company and processes data in the EU by default, which is a strong starting point. For the API, Mistral offers a Data Processing Agreement, commits to hosting in the EU, and does not train on customer data by default. Le Chat (the free consumer chat) collects conversations to improve models unless the user opts out. Whether your use is GDPR-compliant depends on the plan, the DPA and your own documentation: a lawful basis under Art. 6, a record of processing under Art. 30 and, where the risk is high, a DPIA under Art. 35. Not legal advice. As of November 2026.
On Le Chat (the free tier and Pro), Mistral may use conversations to improve its models unless you opt out in the settings. On the API and Enterprise plans, Mistral states that it does not use customer inputs and outputs for training by default. Check the DPA and your plan terms for the exact scope, and switch off optional telemetry if you want zero training use.
Yes. Mistral publishes a Data Processing Agreement for API and Enterprise customers. The DPA covers GDPR obligations including sub-processors, data location and deletion. For Le Chat free accounts there is no business DPA - those run under the consumer terms. Ask your Mistral account manager or check the trust page for the current version.
Mistral AI is headquartered in Paris and processes API requests in the EU. The company has stated that inference for European customers runs on European infrastructure. For Le Chat, the same EU processing applies. This makes Mistral one of the few large-model providers with EU-default hosting, which simplifies the Art. 44 transfer question under GDPR.
Le Chat free is a consumer product without a business DPA. Conversations may be used for model improvement unless the user opts out. For customer or employee data, use the API with a DPA, an Enterprise plan, or a tool like Superkind that wraps the model in a team layer with approval, 1,700+ integrations including Pipedrive, Zendesk and Notion, and its own DPA via the Trust Center.
Superkind uses AI models from OpenAI, Anthropic, Mistral and Google, all processed in Europe. Your team writes to Superkind in Outlook, Microsoft Teams, Slack or the web app, and Superkind does the task in your programs through 1,700+ integrations such as Pipedrive, Zendesk and Notion. It never trains on your data, the DPA is in the Trust Center, and you decide which steps wait for your approval before anything goes out.
Sources
- Mistral AI - Terms of Service and Data Processing Agreement
- Mistral AI - Privacy Policy (data collection, processing location, model training)
- Mistral AI - Le Chat and API product pages (plans and features)
- CNIL - AI and Personal Data: Guidance for Organisations (French data protection authority)
- EUR-Lex - Regulation (EU) 2016/679 (GDPR), Art. 6, 28, 30, 35, 44
- EUR-Lex - Regulation (EU) 2024/1689 (EU AI Act), Art. 4 and Art. 50
- Mistral AI - Trust and Security (sub-processors, hosting, compliance)
- Superkind - Trust Center (DPA, hosting, EU AI Act, sub-processors)
Send Superkind your first task.
Start for free in Outlook, Teams or Slack, hosted in Europe. Or see in 30 minutes how it runs in your systems.

