In accordance with Article 50 of Regulation (EU) 2024/1689 on artificial intelligence (“AI Act”), as amended by Regulation (EU) 2026/1744 known as the “Digital Omnibus”, and Articles 13 and 14 of Regulation (EU) 2016/679 (GDPR).
1. Why this notice exists
Molleva uses artificial intelligence features to help turn content you enter into suggestions that are easier to use.
This notice explains:
- when AI is involved,
- what data is sent and to whom,
- what is not sent,
- your rights and control over these features,
- the system’s limitations.
2. You are interacting with an artificial intelligence system
In accordance with Article 50(1) of Regulation (EU) 2024/1689 (AI Act), Ammar Osei-Mohamend expressly informs you that the features described in this notice involve direct interaction with an artificial intelligence system.
This information is provided in the App before any text is sent for the first time to an AI feature, through a notice displayed the first time you trigger one of them. You can close it without triggering the feature: at that point, nothing has been sent yet.
3. When AI is used
AI is not used all the time. It is involved only when you explicitly trigger one of the following features:
3.1 Breaking down a priority
Molleva can suggest actionable subtasks to help you get started on a priority. You enter the text of your priority, and the AI system generates suggestions that you can accept, edit or reject.
3.2 Voice brain dump
After you dictate what is on your mind, Molleva can prepare a list of items from your dictation. Speech recognition takes place directly on your device (on-device). Only the transcribed text may be sent to the AI system, after you explicitly confirm it.
3.3 OCR import (photo or image)
After text has been extracted from an image or photo, Molleva can prepare a list of items. OCR takes place directly on your device (ML Kit). Only the extracted text may be sent to the AI system, after you explicitly confirm it.
3.4 Date detection and reminder suggestion
When you tap the date detection button for an item in your list, Molleva sends the text of that item to the AI system to look for a date or point in time. If a date is detected, the app shows it to you and asks whether you want to set a reminder. The reminder is scheduled locally only if you explicitly choose “Set reminder”. Only the item’s text is sent to the AI system; the reminder time you choose is then stored on your device.
3.5 Importing pasted text
When you paste a long text, Molleva can turn it into a list of items. Web addresses and email addresses are redacted before transmission: they are removed on the server and never reach the AI system.
3.6 Sorting what you have just cleared
For each item in your incoming list, Molleva can suggest whether it belongs to today or later. Only the text of the relevant items is sent. No item is moved without your confirmation.
3.7 Clarifying a task
Molleva can suggest a clearer rephrasing of an item. Only the text of that item is sent. The suggestion replaces your text only if you accept it.
4. When AI is NOT used
AI is not involved in:
- signing in or creating your account,
- manually writing in your list,
- choosing your daily priorities,
- moving an item to later,
- recording a small win,
- using your local reminders, anchors or habits,
- exporting or deleting your account,
- duplicate detection, which relies on a text similarity comparison performed by the database, without any model,
- speech recognition and text reading from images, which take place on your device,
- most of the App interface.
5. Data sent to the AI system
When you trigger an AI feature, the data sent is limited to:
| Feature | Data sent |
|---|---|
| Breaking down a priority | The text of your priority |
| Voice brain dump | The text transcribed from your voice — not the audio |
| OCR import | The text extracted from the image — not the image |
| Date extraction | The text of the relevant item |
| Importing pasted text | The pasted text, with web addresses and email addresses redacted |
| Sorting the incoming list | The text of the items to sort |
| Clarifying a task | The text of the relevant item |
These texts may contain personal information if you entered it.
Redaction when importing pasted text is applied on the server before any call to the AI system: web addresses and email addresses are removed from the text, and an item containing only that information is discarded.
What is never sent to the AI system:
- your email address,
- your account identifier,
- your authentication data,
- the raw audio of your dictation,
- the raw image of your photo,
- your other data (check-ins, anchors, habits, small wins, intentions).
6. AI system provider
Molleva’s AI system is provided by Mistral AI.
| Information | Value |
|---|---|
| Provider | Mistral AI — French company, RCS Paris 952 418 325 |
| Registered office | 15 rue des Halles, 75001 Paris, France |
| GDPR role | Processor within the meaning of Article 28 of the GDPR |
| Transfer outside the EU | None at provider level — a company established in France and directly subject to the GDPR |
| Contact | privacy@mistral.ai |
Mistral AI is a French company. The processing of your text therefore falls under the European framework applicable to this provider. This information alone does not establish that Mistral’s subprocessors are all based in the European Union: their list and the contractual safeguards must be checked separately.
Molleva makes no independent assurance here about the training on or retention of text by Mistral. These settings depend on the provider’s account, contract and configuration; they must be established and dated before making any more specific promise.
7. What we store — and what we do not store
We do not store:
- the content of your messages sent to AI (the “prompts”),
- the responses generated by AI.
We store only technical usage metadata, strictly necessary to manage quotas and monitor performance:
- AI feature used,
- AI model requested,
- token counts,
- estimated cost,
- request status (success / error),
- latency,
- timestamp.
This metadata is kept for as long as your account exists and deleted when the account is closed.
8. Your control over AI features
You remain in control at every step:
- AI features are entirely optional. All Molleva’s core features work without AI.
- You explicitly confirm each transmission to the AI — no text is sent automatically.
- You can review, edit or reject every suggestion before using it.
- You can choose not to confirm — nothing is added to the App without your confirmation.
- Usage quotas apply according to your plan (free or Premium).
9. Limitations of the AI system
The AI system may:
- misunderstand a sentence or intention,
- suggest subtasks that are too vague or unsuitable,
- produce wording that does not match what you meant,
- be temporarily unavailable,
- refuse a request,
- hit a usage limit (hourly, daily, weekly or monthly, depending on the feature),
- produce an inaccurate or incorrect response (“hallucination”).
Always check suggestions before using them.
10. What Molleva’s AI does not do
Molleva’s AI system does not provide:
- medical or psychological diagnoses,
- therapy or mental health support,
- legal, financial or tax advice,
- decisions that produce legal effects for you,
- emergency or distress response.
Molleva is not a medical or therapeutic app. If you are in distress, contact a professional or call 3114 (the national suicide prevention number — free, 24 hours a day).
11. AI Act classification — Limited risk
Molleva’s AI system is classified as limited risk within the meaning of Article 6 and Annex III of Regulation (EU) 2024/1689 (AI Act), as amended by Regulation (EU) 2026/1744 (“Digital Omnibus”), which entered into force on 27 July 2026.
This amending regulation postponed the obligations applicable to high-risk systems until 2 December 2027 and granted an extension until 2 December 2026 for the marking required under Article 50(2). The transparency obligations in Article 50(1) — telling you that you are interacting with AI — have remained applicable since 2 August 2026, and Molleva complies with them.
- It makes no automated decisions that produce legal or significant effects for you (Art. 22 GDPR).
- It does not fall into the high-risk AI categories (no biometrics, no scoring, no administrative decision).
- It does not generate content that could be mistaken for real content (no deepfakes, no image generation).
12. Marking generated content
Article 50(2) of the AI Act requires content produced by an AI system to be identifiable. Molleva applies this marking at the level of your data, and not only on screen.
On screen. Each preview screen for an AI feature displays the notice “AI-generated suggestions — review before using.” when you decide whether to accept the suggestions. The same screen displays the action “Something wrong with this result?” so you can report an unsuitable output.
In your data. The following is retained:
| What is marked | What is recorded |
|---|---|
| Each item you capture | Its source: typed manually, dictated, photographed or pasted text |
| Each subtask created by breaking down a priority | That it was suggested by a model, which model, and whether you edited it before confirming it |
| Each text produced through clarification | That it came from AI-assisted rephrasing, and which model |
If you rewrite a text from a clarification yourself, the marking is removed: the content is then no longer AI-assisted in origin.
These markers appear in your data export.
Timing. The marking requirement in Article 50(2) applies no later than 2 December 2026, under the extension granted by Regulation (EU) 2026/1744. Molleva applies it in advance.
13. Potentially sensitive data
Molleva does not ask you to enter sensitive information (health, sexual orientation, religious or political beliefs, biometric data).
However, you can freely enter whatever is on your mind. If sensitive content is included in text sent to an AI feature, it will be sent to Mistral AI under the conditions described above.
We encourage you not to send AI information that you do not want processed by an external provider.
14. Your GDPR rights over data processed by AI
You can exercise your rights of access, rectification, erasure, restriction, objection and portability as described in our Privacy Policy.
Since prompts and responses are not stored, it is not possible to provide or retroactively erase the content of a past request.
The technical usage metadata described in §7 is not included in the export file. It remains covered by your right of access: you can request it at the address below, and it is deleted with your account.
Your export does include the following AI-related information: the source of each item you captured (typed manually, dictated, photographed, pasted text), and the marking of each subtask created by breaking down a priority — indicating that it was suggested by a model, which model, and whether you edited it before confirming it.
Data protection contact: contact@molleva.fr
CNIL complaint: www.cnil.fr — 3 place de Fontenoy, TSA 80715, 75334 Paris Cedex 07.
15. Updates to this notice
This notice will be updated in the event of:
- a change of AI provider,
- the addition of a new AI feature,
- changes to applicable legal requirements (AI Act, GDPR).
You will be informed in the App of any substantial change.