Explicit Consent Text
Operations that need no consent
Your account, goals, check-ins, subscription and the transfer abroad for AI assessment rely not on explicit consent but on performance of a contract (KVKK Art. 5(2)(c), GDPR Art. 6(1)(b)). No consent is asked for these.
Why not consent: Demostra's core undertaking is to assess your explanation, and that assessment cannot be produced without sending text to the AI provider. The transfer is therefore not an option within the service, it is the service. Tying such an operation to consent would be wrong twice over:
- The consent would be invalid. KVKK requires explicit consent to be given by free will; conditioning access to a service on consent vitiates it. The Board established exactly this against a private health institution in decision 2023/692 of 2 May 2023 and imposed sanctions.
- It would be worse for you. An invalid consent means the transfer has no legal basis at all, worse than never asking.
So the transfer is disclosed as part of the Terms of Service. If you do not accept the Terms, no account is created, that is not forced consent but the ordinary consequence of no contract being formed. We hide nothing about it: Privacy Notice §3 names the transfer, the recipient, the country and the risks.
The appropriate safeguard route for this transfer, a standard contract, requires the provider's signature, and that signature has not been obtained yet. Nor is there any country for which the Board has issued an adequacy decision. Detail: Privacy Notice §3.
Operations that need consent
Sharing dictation corrections
When you correct a term the speech recogniser misheard, you may agree to share only the misheard phrase → correct term pair. The sentence you spoke is not sent.
This is genuinely optional: it is not needed for the service to work, which is why it can rest on consent and is off by default. Refusing it affects no feature of the app.
It stays linked to your account for 30 days; if you withdraw consent within that window or delete your account, the contributions are deleted too. Withdraw at Settings → Data.
Cloud refinement
The text you see while speaking is always transcribed on your device. That does not change, cannot be switched off, and is not an option: it is how the product works.
In Settings, you can consent to turn on cloud refinement. The only thing it changes is what happens after you finish: the recording is uploaded to our server and passed to OpenAI in the United States, and the more accurate transcript that comes back replaces the one your device produced. The difference is largest on medical and engineering vocabulary.
The audio is never stored: it is not written to disk on our server, not saved to a database, not written to any log. It is discarded as soon as it is transcribed. The temporary recording on your device is deleted too, whether or not the upload succeeded.
This too is genuinely optional and off by default. Refusing it, or simply never turning it on, affects no feature of the app, your transcript just stays as your device produced it. You can withdraw it in Settings at any time; from your next recording onward your audio is not sent at all.
Risks you should know about. This is a transfer abroad: your audio may be processed on OpenAI's servers outside Türkiye. We cannot guarantee that the data is protected there to the same standard as under Turkish data protection law, because that processing follows their rules, not ours. If something goes wrong, your means of seeking redress may also differ from those at home. You give this consent knowing that; refusing it or withdrawing it later carries no consequence whatsoever.
About what you say. We do not choose your subject matter: if your goal is medicine you may be describing diseases, treatments or cases; if you study law, religion or politics your speech may touch those areas. Turkish data protection law treats such data as special categories and protects it more strictly. With cloud refinement on, your audio travels with whatever it contains, and this consent covers that too. If you would rather your voice never left the device while explaining such a topic, leaving the switch off is enough: on-device recognition keeps working for every subject.
Consent is not a condition of service
KVKK requires explicit consent to be given by free will. Neither consent on this page (dictation corrections, cloud refinement) is a condition of any feature. Accordingly:
- Refusing either or both of these consents affects neither your account nor any feature.
- Withdrawing consent does not close your account.
- You are not asked to approve having read the privacy notice.
- Accepting the Terms of Service is required, but that is contract formation, not a consent.
Scope of consent
The consent you give is limited to the purposes listed above. Data transferred under it:
- Is not used for advertising or marketing
- Is not sold to third parties or passed to data brokers
- For dictation corrections sent to our AI provider: not used to train the provider's models, under the provider's published policy, API inputs and outputs are not used for model training by default
- For cloud refinement: the recording goes to our server first and then to OpenAI. We are the controller of that transfer and OpenAI processes it on our behalf, the audio is never stored at any stage
- If a new purpose arises, consent is asked again and separately, no “blanket consent” is taken
Contact
info@demostra.app · Privacy Notice · Terms of Service · Refund policy · Delete account