← All pages

1. What the document is: Mistral AI's consumer Terms of Service for EU/EEA users of Le Chat and Mistral AI Studio — explicitly split from a separate "Consumer Terms for the rest of the world" and from the Commercial Terms (business customers).

What this wiki found — complete, every page

Presence facts, not verdicts — each flagged term links to its definition and the exact clause on this page. Absence of a badge means the term isn't currently flagged here, not that the page is risk-free.

Le Chat – Terms of Service

Dataset: GenGA (Generative AI Governance Archive) — 11 AI services, 2025–present

GenGA (Generative AI Governance Archive) is this wiki’s Generative-AI dataset: raw policy snapshots captured by the Open Terms Archive genai-eu project across 11 GenAI providers (ChatGPT, Claude.ai, DeepSeek, Google Generative AI Services, Le Chat, Llama API, Meta AI, Microsoft Copilot, Perplexity, Qwen Chat, xAI). Unlike PGAv2, GenGA has no pre-tagged risk scores — all risk scoring and keyword tagging on these pages is LLM-assigned by direct reading, spanning 2025–present.

Note: This page contains documented policy clauses. Risk assessment is qualitative and context-dependent. For analysis of patterns across platforms, see:

• AI Act Substance Matrix
• Tier Bifurcation
• Scraper Enforcement Gap
• Methodology

Source: sources/GenGA/Le Chat/Terms of Service/ (raw Markdown captures, Open Terms Archive genai-eu dataset — no pre-tagged JSONL exists for this source) Last updated: 2026-04-16

Note on methodology: No pre-computed risk_score/keyword fields exist for this source. Keyword tags and clause analysis on this page are LLM-assigned, using the same rubric/vocabulary as the PGAv2 pages plus new GenAI-specific tags (output ownership, competing model ban, feedback, eu ai act). Treat as first-pass analysis, not externally verified ground truth. This wiki’s risk-scoring system was retired project-wide on 2026-06-21 (see methodology.md §4) — no page, GenGA included, computes or displays a numeric risk score.

Skipped capture: 2025-12-15T11-34-46Z.md (31 bytes, “Redirecting to legal.mistral.ai”) is a failed scrape, not a real policy version — skipped.

Scraper-artifact note: A large number of captures between 2026-01-27 and 2026-03-10 differ from one another only in a rotating CDN deployment hash (dpl=...) embedded in the Mistral AI logo image URL — the same kind of cosmetic CDN-link-rotation artifact documented elsewhere in this wiki. These are collapsed into a single date-range row below per the wiki’s unchanged-date convention; an individual ### heading is still emitted for each raw snapshot.

Overview

1. What the document is: Mistral AI’s consumer Terms of Service for EU/EEA users of Le Chat and Mistral AI Studio — explicitly split from a separate “Consumer Terms for the rest of the world” and from the Commercial Terms (business customers).

2. Input rights: Same opt-out-by-default structure as Le Chat’s Commercial Terms and Privacy Policy: training-eligible Input/Output unless on Le Chat Enterprise/paid API, or (depending on capture date) Le Chat Pro/Student, subject to opt-out. A “Fine-Tuned Model” clause clarifies fine-tuned models are not “Output” and aren’t used by Mistral except to serve them back to the customer.

3. Output restrictions: Same narrow image-Output competing model ban as the Commercial Terms; a parallel restriction bans using Third-Party (web-search) Content for machine-learning purposes or to build competing commercial products.

4. Non-explicit predatory clauses: A “Conversation Links” feature lets users share conversations via public links with no access monitoring or control by Mistral (“we do not monitor or control who accesses your conversations”). A 2026-03-26 addition (status fluctuated across nearby captures — see Changes Summary) introduces voice-cloning (“Audio Mistral AI Products”) with a built-in AI-disclosure requirement.

5. Regulatory references: Major, unique finding: the 2025-11-11 baseline included a full “Additional Terms for AI Laws Responsibilities” annex — the most detailed EU AI Act citation found anywhere in this dataset, defining “Applicable AI Laws” by reference to “Regulation (EU) 2024/1689… (the ‘AI Act’),” the AI Office, the European Artificial Intelligence Board, GPAI “systemic risk” classification, Article 53.1(b)/53.2 AI Act documentation obligations, and a 48-hour Serious Incident reporting duty. This entire annex was removed by the 2025-12-15 rewrite and has not reappeared in any later capture.

6. Regional variation: This document is itself the EU/EEA-specific consumer ToS (explicitly distinguished from a “Consumer Terms for the rest of the world” referenced but not captured in this dataset). A Belgium-specific digital-services conformity guarantee was added 2026-01-27.

7. Key risk to users: The EU AI Act annex removal is the standout finding — Mistral briefly had the most detailed GPAI-systemic-risk compliance framework in this dataset, then withdrew it from the consumer-facing Terms within about a month, with no equivalent reappearing since.

Flagged Keywords & Risks (LLM-assigned)

• royalty-free, train AI/models — training happens by default (opt-out) under a royalty-free license, and the tier exceptions change in step with the Commercial Terms and Privacy Policy. Why it matters: confirms the same trend seen across Le Chat’s other contracts — a no-training promise eroding as more tiers become training-eligible.
• competing model ban — narrow image-Output and web-search-Output restrictions. Why it matters: same scope as the Commercial Terms’ narrower (image-only) ban, contrasting with OpenAI’s/Anthropic’s all-Output bans.
• feedback — Feedback usage is tied to training, but only “to the extent strictly necessary.” Why it matters: narrower than the Commercial Terms’ unconditional Feedback grant, since it adds a necessity qualifier.
• output ownership — user retains Input ownership and owns all Output; Mistral assigns any Output rights it might hold to the user. Why it matters: consistent, user-protective pattern across this provider’s documents.
• eu ai act — the now-removed “Additional Terms for AI Laws Responsibilities” annex. Why it matters: the only confirmed EU AI Act citation found anywhere in this 11-provider dataset, making its removal a significant negative data point for regulatory-transparency trend-tracking.

Regulatory & Research Context

Applying Davidson et al.’s (2026) “regulatory gray areas” framework by analogy (Mistral/Le Chat was not among the five providers they directly studied), the removal of this page’s EU AI Act annex just over a month after baseline — taking with it the GPAI systemic-risk classification, Article 53 documentation duties, and 48-hour Serious Incident reporting duty — converts what had been the dataset’s single most explicit AI-specific regulatory commitment into exactly the kind of uncertainty their framework describes. Pandit et al. (2026), who directly studied Mistral/Le Chat’s terms, separately found that “all terms in our analysis contained language that explicitly discards assurances regarding the quality, availability and appropriateness of the service,” a pattern consistent with a provider narrowing its own compliance commitments over time rather than expanding them.

Changes Summary

Version History

2025-11-11

• Explicit AI clause: YES
• Non-explicit predatory: YES
• Flagged keywords: royalty-free, train AI/models, competing model ban, feedback, output ownership, eu ai act

Clause: output ownership

Your Data. You and your authorized users may provide input to the Mistral AI Products, such as prompts or fine-tuning data (“Input”), and receive output from the Mistral AI Products based on such Input (“Output”). […] To the extent permitted by applicable law, you (i) retain all ownership rights in Input and (ii) own all Output. We assign to you all right, title, and interest, if any, in and to Output that we may have.

Clause: competing model ban

Output restrictions. To the extent permitted by the applicable laws, you may not use: (a) image Outputs to develop or train any competing image-generation Mistral AI Product and (b) Outputs generated using our web-search feature to (i) copy, store, archive, cache or create a database of the Outputs, (ii) redistribute, resell, or sublicense the Outputs, (iii) as part of any machine learning or similar algorithmic activity, or (iv) to create, train, evaluate or improve Your Offerings.

Clause: feedback, train AI/models

Feedback. For certain Mistral AI Products, such as Le Chat, you have the option to provide feedback by selecting “thumbs up” or “thumbs down” (“Feedback”). By providing Feedback, you acknowledge and agree that Mistral AI may use your Feedback, along with the associated Input and Output data, for the purpose of training our artificial intelligence models and improving the Mistral AI Products.

Clause: royalty-free, train AI/models

How we use Your Data. We do not use Your Data to train our artificial intelligence models except (a) when you use the Mistral AI Products under a free subscription, Le Chat Pro or Le Chat Student and you have not opted-out of training, (b) when you use Le Chat and provide Feedback to us, and (c) when Your Data is flagged as part of our automated moderation […] You grant us a worldwide, non-exclusive, non-transferable […] royalty-free, fully-paid license […] to use Your Data for the purpose of (a) providing, maintaining and optimizing the Mistral AI Products […] but excludes model training, (b) performing our obligations under these Terms and (c) as applicable and solely in the cases provided in the previous sentence, to train our artificial intelligence models.

Clause: eu ai act (removed 2025-12-15)

“Applicable AI Laws” means any applicable artificial intelligence law or regulation, including, to the extent applicable (a) Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonized rules on artificial intelligence including any associated implementation or delegated acts (the “AI Act”), and (b) mandatory guidelines published by a Competent Authority. […] Customer acknowledges that Mistral AI is responsible for reporting to the Competent Authority any Serious Incident relating to any Covered Mistral AI Product classified as a General-Purpose AI Model with systemic risks. Accordingly, Customer shall notify Mistral AI of any Serious Incident within forty-eight (48) hours of becoming aware of such Serious Incident […]

2025-12-15

output ownership/competing model ban/feedback/train AI/models/royalty-free unchanged from 2025-11-11 (re-stated with the same substantive terms, “Le Chat Pro or Le Chat Student” tier wording retained).

Clause: eu ai act — REMOVED

[The “Additional Terms for AI Laws Responsibilities” annex quoted above does not appear anywhere in this or any later capture in this dataset.]

Clause: EU/rest-of-world split (new, not separately concept-tagged)

These Terms of Service (these “Terms”) govern your personal use as an EU user of Mistral AI Studio, Le Chat, Mistral Code, and the other websites, products, services, and technologies we offer (the “Mistral AI Products”). If you are a consumer residing outside the European Union, the Consumer Terms for the rest of the world apply to you.

2026-01-26

output ownership/competing model ban/feedback/train AI/models/royalty-free unchanged from 2025-11-11.

Clause: Data Switching (new, not separately concept-tagged — user-protective)

Switching. You have the right to switch to another service provider or your own infrastructure. […] Exportable data and digital assets. You may export your data and digital assets via the export features provided on your Mistral AI Account, which includes at least your Le Chat conversations (prompts and outputs).

2026-01-27

All flagged clauses unchanged from 2026-01-26.

Clause: Belgium conformity guarantee (new, not separately concept-tagged)

Consumers residing in Belgium also benefit from the following legal guarantee for digital services, in accordance with Articles 1701/7 and following of the former Belgian Civil Code.

2026-01-27 (23:34) through 2026-03-10

All clauses unchanged from 2026-01-27 across this entire date range (2026-01-27T23-34-56Z, 2026-02-25T23-33-40Z, 2026-02-26T11-33-35Z, 2026-02-28T11-33-23Z, 2026-03-01T11-33-04Z, 2026-03-02T11-33-12Z, 2026-03-03T11-33-37Z, 2026-03-03T23-33-39Z, 2026-03-04T23-33-44Z, 2026-03-05T11-33-22Z, 2026-03-05T23-33-01Z, 2026-03-06T11-33-05Z, 2026-03-09T11-33-06Z, 2026-03-09T23-34-06Z, 2026-03-10T23-35-06Z) — each transition differs only in the Mistral AI logo image’s rotating CDN deployment hash (dpl=... query parameter), a cosmetic scraper artifact.

2026-03-12

output ownership/competing model ban/feedback/train AI/models/royalty-free unchanged from 2025-11-11.

Clause: VAT/pricing disclosure (new, not separately concept-tagged)

Price. The applicable pricing (including any applicable taxes) for each paid Mistral AI Product, and the accepted means of payment are indicated on our pricing page. The prices of the Mistral AI Products are indicated “all taxes included”, with Value Added Tax (VAT) applied at the rate in force in your billing country on the date you subscribe to a Mistral AI Product.

2026-03-26

All flagged clauses unchanged from 2026-03-12.

Clause: Audio Mistral AI Products / ai disclosure (new — presence fluctuated in subsequent captures, see Changes Summary)

Audio Mistral AI Products. We may provide Mistral AI Products such as models or APIs capable of generating audio outputs, including through voice cloning features. […] You must disclose AI-generated or partially AI-generated content generated through the audio Mistral AI Product where required by applicable law.

2026-04-02

All flagged clauses unchanged from 2026-03-12 — cosmetic-only capture; the Audio Mistral AI Products clause from 2026-03-26 is transiently absent here.

2026-04-16

All flagged clauses unchanged from 2026-03-12; the Audio Mistral AI Products clause is present again, unchanged in wording from 2026-03-26.