Privacy Policy

• We collect what we need to run your study plan, practice and coaching, and nothing for advertising.
• We don’t sell your data, and we don’t use your private notes, answers or conversations to train AI models.
• When you use an AI feature, the text needed for that request is sent to an AI model to produce your answer. Section 5 explains exactly which providers are involved.
• You can export your cards, notes and plans and delete your account yourself in Settings, at any time.
• Strictly necessary cookies keep FluxStudy working. Optional preference and analytics cookies are only used if you allow them, and you can withdraw at any time in Cookie settings.
• Your EU/EEA and UK data-protection rights are summarised on our GDPR page.

The table below lists each category of personal data, what it is used for, the legal basis under Article 6 of the GDPR, how long it is kept and who processes it for us. The legal bases we rely on are:

• Contract (Art. 6(1)(b)): processing needed to give you the service you signed up for.
• Legitimate interests (Art. 6(1)(f)): for example keeping FluxStudy secure and reliable. We have balanced these interests against your rights, and you can object (see section 11).
• Legal obligation (Art. 6(1)(c)): for example keeping accounting records.
• Consent (Art. 6(1)(a)): optional preference and analytics cookies, which under the ePrivacy rules (in Sweden, the Electronic Communications Act, LEK) need your consent before anything is stored on or read from your device. You can withdraw it at any time.

Some data is kept only on your device, such as offline flashcards, unsaved mock-exam answers and your sound setting. See our Cookie & Storage Policy for the full list.

• From you: your email address, what you upload, type, answer and choose.
• From your use of FluxStudy: session records, product events, progress calculated from your practice.
• From Stripe: subscription status and renewal details if you buy a paid plan.

You can build a study plan before you create an account. Your answers (first name, goal, exam date, your reason, subjects and any notes you paste) are kept in your browser (or the app) on your device, so you can pick up where you left off, for up to 7 days.

To show your plan and sample practice cards, our servers keep a temporary setup record with only what those previews need: your goal, exam date, subjects, study times, language and time zone, the notes you added (deleted 24 hours after they were last used), the dates found in them and the sample cards. It never holds your name, your reason, your email address or your IP address. If you add notes, the sample cards are written by an AI model as described in the next section. The record is deleted 7 days after your last change (and never kept longer than 14 days), or right away when you choose “Start over”. When you create your account, the answers are moved into it and the temporary record is emptied at once; an empty reference is kept for 24 hours so a repeated save can’t create duplicates. Legal basis: taking steps at your request before an account exists (Article 6(1)(b) GDPR) and our legitimate interest in preventing abuse (Article 6(1)(f)).

To stop automated abuse, creating a setup record needs a bot check, and a strictly necessary cookie (flux-onb-dev, a random value, 30 days) counts how many setups and sample cards one browser starts. Usage limits by network use a one-way hash, never the IP address itself.

Flux, practice generation, mock exams and study images are AI features. When you use one, we send the content needed for that request to an AI model and return the result to you. That content can be your message, relevant passages from your own materials and notes, and preferences you asked Flux to remember. Specifically:

• Reading your materials: text extraction from files and search embeddings run on Cloudflare Workers AI; photos and scans are read as described under “Camera and photos” below.
• Searching your library: Flux can search your own notes and materials (one note, one course or your whole library, as you choose) and notes other learners explicitly shared with you, which are labelled as shared. Search embeddings of your notes and of your question are made on Cloudflare Workers AI and stored in Cloudflare Vectorize; only the few most relevant passages are sent with your message, and Flux shows which ones it used. Flux never searches other people’s content that wasn’t shared with you. Deleting or trashing a note removes its passages and embeddings from search.
• Flux coaching and deep tutoring are sent through OpenRouter. Jev selects GPT-6 Luna for simpler chat and GPT-6 Luna Pro for more complex chat; Deep mode uses Pro. Those requests run on Microsoft Azure with zero data retention requested and no use for training. If the route is busy or unavailable, Flux uses the providers below. Practice questions, mock exams and syllabus reading, and fallback coaching requests, use DeepSeek V4 Flash models through OpenRouter. OpenRouter selects zero-data-retention providers such as DeepInfra, Fireworks and Together, outside China, and never DeepSeek’s own service. If those routes fail, Cloudflare Workers AI answers instead (currently GLM-5.3-Flash).
• Safety and routing checks: each message you send Flux (and, in a live voice session, the transcript of each thing you say) is also checked by TypeSafe’s Jev model, reached through OpenRouter: for signs that you may be in crisis or in an emergency (then Flux stops and gives you a crisis line), whether you are asking Flux to do something in the app, how complex your chat request is, and, during a mock exam, whether you are asking for help that would break exam conditions. Jev only answers yes/no-style questions; it writes nothing. The same check runs on the material you paste in the onboarding preview. Our own keyword rules always run as well, and if the check is slow or unavailable they decide alone. We keep only counts of these checks, never the message.
• Checking typed answers (when available): a typed answer is compared with the card’s answer key, first by our own code, then by TypeSafe’s Jev model and, when Jev isn’t sure, by the text model described above. Only the question, the answer key and your answer are sent. If none of them is sure, you check the answer yourself and it isn’t counted against your allowance.
• Study images are created from the description you write on Cloudflare Workers AI (currently the FLUX.2 [klein] model). We may instead send the description through OpenRouter to xAI’s image model. Before an image is shown, the description and the image are checked for unsafe content by Cloudflare Workers AI models (Llama Guard and GLM-5.3-Flash); images that fail the check are not shown or kept. Your uploaded photos are not sent to this route. The resulting image is stored privately in your account.

Uploaded content is treated as data, not instructions: text inside a document can’t make Flux save preferences or act for you. We don’t use your content to train AI models. The providers’ own handling of requests is governed by their terms and our agreements with them. Content sent through OpenRouter may be processed outside the EU, mostly in the United States (see section 9). Messages in People chats (friends, groups and classes) are never sent to any AI model, these checks included.

Camera and photos

Flux, Notes and the Library can read a photographed problem, classroom board or handout. The camera only turns on after you tap “Open camera” in one of these tools, and it turns off when you close it; you can always pick an existing photo instead. Your browser asks for permission first, and no embedded third party can use the camera.

• Photos are resized and re-encoded on your device before upload, which removes location and camera details (EXIF).
• They are sent only to read the text, maths or layout on them: first through OpenRouter to a zero-data-retention inference provider (currently DeepSeek V4.1 Flash, as above), and if that fails to Cloudflare Workers AI (currently GLM-5.3-Flash, with Meta’s Llama 4 Scout as a last resort). Each photo read counts toward your scanned-pages allowance.
• A photo you solve with Flux isn’t stored unless you choose “Save to notes”, which keeps it privately as a note image.
• A scanned handout keeps its page photos privately with the material you save, next to the text you reviewed, so you can look at the actual pages later. You can delete one photo or all of them at any time in the Library (the text stays); deleting the material or your account deletes them too.
• A board or paper scan keeps the photos in the note by default (“Keep the photos in the note”); you can switch that off before saving or remove the images from the note later.
• Stored photos count toward your storage, are only shown to you, and are included in your data export.
• Photos are never used to recognise or identify people; faces in them are ignored. We don’t use your photos to train AI models.

AI can be wrong

AI output can contain mistakes. Practice scores and progress states are study aids, not grades or predictions of exam results. You can report any item you think is wrong.

No automated decisions with legal effects

FluxStudy does not make decisions based solely on automated processing that have legal or similarly significant effects on you (Art. 22 GDPR). Review schedules and progress states only organise your own study, and you stay in control of your plan.

Wherever you write in FluxStudy (answers, cards, notes, chats, plans, search) you can press the microphone button and speak instead of typing; the Flux chat’s microphone works the same way. Nothing is recorded until you press or hold the button, and a clip stops after one minute.

• The clip is sent to Cloudflare Workers AI (Whisper speech-to-text), which processes it only to return the text. FluxStudy does not store the audio.
• The text is only put into the field you were writing in, where you can change it before anything is sent. From there it is handled exactly like text you typed (for example, a note you save or a message you send).
• We keep a count of the seconds you used, to apply your plan’s monthly voice-typing minutes and to see what the feature costs us — not what you said.

In a live voice session (Flux → Live) your microphone audio is processed by OpenAI or Google (Gemini), depending on the configured session service. The provider runs the realtime speech model that hears you and answers with a voice. Before your first session we show you how voice is processed and record your agreement, and your browser or device asks for microphone permission. You can take that agreement back at any time in Settings → Privacy & data. The microphone is only used while a session is running.

• No audio is stored by FluxStudy. OpenAI processes the audio to run the session and may keep API data for up to 30 days to detect and prevent abuse, after which it is deleted. OpenAI does not use API data to train its models. Google’s paid API terms do not permit using prompts or responses to improve its products, but Google logs them for a limited period for abuse monitoring. We do not enable Gemini session-resumption storage; fresh connections receive only a short text history and relevant retrieved context.
• The speech provider also transcribes what is said, so Flux can follow the conversation and we can show live captions. For oral exam and thesis defense practice we keep the written transcript for 90 days to produce and show your feedback report; you can delete it at any time. The report is written by our text AI providers (OpenRouter and the zero-data-retention inference providers it routes to, or Cloudflare Workers AI; see section 9).
• The transcript of each thing you say is checked for signs of crisis or an emergency by our keyword rules and by TypeSafe’s Jev model (through OpenRouter, see section 5); if either fires, Flux stops and gives you a crisis line. Only counts of these checks are kept.
• For tutoring, study pal and coaching sessions we keep no transcript. It is held for up to 30 minutes after the call only so you can choose “Save to notes”; then it is deleted.
• We keep session details (mode, language, length, why it ended, minutes billed and an estimated AI cost) and your voice minute ledger to bill minutes correctly. We send OpenAI a pseudonymous identifier (a one-way hash, not your account id or email) so abuse can be traced without identifying you.
• If you buy minutes in the iOS app and ask Apple for a refund, Apple may ask us how much of the purchase you used. We only answer if you allowed it in the app; otherwise we don’t share anything.
• These providers process data in the United States and other countries (see section 9). Flux’s feedback is not an official grade.

In song mode Flux writes lyrics from your own material, and Google’s Lyria music model, reached through OpenRouter, turns them into a short song. Only the lyrics and the style you pick (genre, mood, voice, tempo) are sent — never your name, email or account id.

• Before anything is sent, every version of the lyrics is checked for safety by Cloudflare Workers AI and our text AI providers (see section 9).
• We ask the route not to log the request or train on it (data_collection: "deny"). Unlike the text models in section 9, the music model runs on a single provider that is not zero-retention, so Google may still keep the lyrics and the finished song for a limited period to detect abuse.
• The finished song (MP3) is stored in our file storage with its lyrics and line timings until you delete it or your account. To show synced lyrics we transcribe the song with Cloudflare Workers AI.
• Listening to a song doesn’t change your progress. Your answers to the quiz after a song are kept as practice, like other practice answers.
• Google processes this data in the United States and other countries (see section 9).

We use a small number of providers to run FluxStudy. Each may only process personal data for the purposes described here. We don’t sell personal data or share it with advertisers.

AppHelm is a product of our own corporate group, so analytics data stays within Sansa Group AB rather than going to an unrelated company.

Data processed worldwide

Our vendors for hosting, payments, AI and email may store and process your data on servers around the world, including outside the EU/EEA and the UK. Where the law requires it, these vendors rely on recognised safeguards such as the EU Standard Contractual Clauses or the EU–US Data Privacy Framework. You accept this Privacy Policy and our Terms of Service when you sign in or use FluxStudy. If you don’t agree to your data being processed this way, please don’t use FluxStudy.

We may also disclose data where the law requires it, to protect people’s safety or our legal rights, or as part of a merger or sale of the business, with this policy continuing to apply.

Your data is kept while your account is active and deleted when you delete your account. The retention column in section 3 gives the specifics. When you delete your account from Settings, this is what happens:

• All your sessions are ended immediately, on every device.
• Your uploaded files and generated images are deleted from storage, and your search embeddings are deleted.
• Your account and study data are deleted from our database: materials and extracted text, courses, goals, plans, practice items, answers, review schedule, mock exams, focus sessions, rewards, Flux preferences, reminder settings, product events and usage records.
• A live voice session that is still running is ended, then your voice sessions, transcripts, feedback reports and voice consent are deleted. Records of voice minute purchases and usage are kept for accounting, disconnected from you.
• Content reports you filed are kept for quality and abuse-prevention records, but disconnected from you.
• Billing event receipts are kept as accounting law requires, and Stripe keeps its own transaction records. Deleting your FluxStudy account does not by itself cancel a paid subscription, so please cancel it first (Pricing → “Current plan · manage”).
• Copies in our database provider’s point-in-time recovery backups expire on a rolling basis and are not used to restore deleted accounts.

Deleting your account also removes some storage on the device you use at the time. To remove everything from a device, clear site data for fluxstudy.com (see the Cookie & Storage Policy).

Under the GDPR you have the right to:

• Access the personal data we hold about you and get a copy.
• Rectification: correct inaccurate data. You can change most things yourself in the app.
• Erasure: have your data deleted. Use “Delete account” in Settings → Account, or ask us.
• Restriction: ask us to limit processing while an issue is resolved.
• Data portability: receive your data in a machine-readable format. In Settings you can export flashcards (CSV), notes (Markdown) and plans (CSV), and your committed plan as a calendar file (ICS). Ask us for anything else.
• Object to processing based on legitimate interests, such as product events or reminders. You can switch reminders off in Settings.
• Withdraw consent at any time where we rely on consent (optional preference and analytics cookies). Use “Cookie settings” in the site footer, the Cookie & Storage Policy or Settings → Privacy & data. Withdrawing is as easy as giving consent and doesn’t affect processing that happened before. FluxStudy’s core service never depends on consent to optional cookies.
• Complain to a supervisory authority: in Sweden, Integritetsskyddsmyndigheten (IMY), imy.se (opens in a new tab), or the authority where you live or work. We’d appreciate the chance to help first.

Our GDPR page explains each right, the legal bases we use and how international transfers are protected in more detail.

Study buddies are off until you invite someone or accept an invite, and you can have at most 5. A buddy sees only your first name, whether you studied today (yes or no), how many days you studied this week inside a shared weekly goal you both joined, and your streak count if you turn that on. You choose this per buddy in Progress. Buddies never see your notes, cards, grades, mock scores, minutes, times of day, courses, exam dates or your “why”. There are no rankings, no public profiles and no notifications.

Invite links work once and expire after 7 days; we store only a scrambled (hashed) version of the link. If you invite someone by email, we send that one invite and keep only a hashed form of the address, to limit invites to 5 a day. Removing a buddy is instant and they aren’t told. Expired invites are deleted daily, cheers after 30 days and shared goals after 12 weeks. Buddy data is included in your export and deleted with your account (legal basis: contract — 6(1)(b)).

Any learner can create a class and teach it, or join one with an invite. A teacher sees your first name and that you joined. Nothing about your studying is shared by default. In each class you can choose to share study days per week, practice on the class’s pack cards, progress on the pack’s objectives and mock scores on pack courses, and turn each off again at any time; we record each choice (consent, 6(1)(a)). Teachers never see your notes, Flux chats, your “why”, habits, results outside the class’s packs or times finer than a day. Class averages only appear when at least 5 students share the same thing, and there are no rankings.

Course packs a teacher publishes are copied into your own library when you add them, and stay yours if you leave. Files a teacher shares stay in the teacher’s storage and are viewable only by class members while the pack is live; teachers confirm they have the right to share them. Invite links expire and are stored only as a hash; email invites keep only a hashed address for rate limiting, and expired invites are deleted daily. Leaving a class is instant. Class data is in your export. If you delete your account, your memberships and sharing choices are deleted; a class you taught alone is closed (students keep their copies), and packs you created in a class with other teachers are removed (legal basis for running classes: contract, 6(1)(b)).

People is off until you add a friend, accept an invite or join a group. Friends and group members see your first name and what you choose to write or share with them. They never see your email address, your study data, grades or activity (study buddies keep their own rules, above). You can add a friend with your friend code, an invite link or their email address; we never tell anyone whether an address has an account. If you invite an address that has no account, we send that one invite and keep only a scrambled (hashed) form of the address.

In a teacher-led class, chat is off unless a teacher turns it on. When it’s on, everyone in the class chat sees each other’s first names and messages (never email addresses or study data). Anyone, teachers included, can opt out: you’re then hidden from the chat, your messages are hidden from the others, and you get no notifications from it.

You can also share files (PDFs, pictures, Word and PowerPoint files; a Beta feature) with friends and groups. You must confirm that you own a file or have the right to share it, and we keep a record of that confirmation (who, when and which wording) while the share is active and for at least 12 months after. A shared file stays in your storage and counts toward your storage allowance, not the recipients’; they can view and download it. Recipients can report a file for copyright, and we remove files that break the rules.

Messages, and notes, card decks or files you share, are stored on Cloudflare’s servers with the rest of your data. We keep a message until you or the other person deletes it, the conversation or group is deleted, or 12 months after it was sent, whichever comes first: a daily cleanup deletes messages older than 12 months. Deleting a message removes its text at once. In the iOS app, your chat list and the latest messages of chats you opened recently are kept on your device so you can read them offline; they’re removed when you sign out. Answered friend requests are deleted after 30 days and expired invite links daily. Nobody at FluxStudy reads your messages unless someone in the chat reports one (see below), and no message is sent to any AI model, used for training or put in a notification or email: notifications only say who wrote. If you report a message, a copy of that one message goes into the report so a moderator can review it; reports are kept like other reports.

You can block or report anyone you’re connected to, mute a chat, leave a group and stop sharing at any time. People data is in your export. When you delete your account, your messages, direct chats, friendships, requests and shares are deleted; a group you own passes to its longest-serving admin or member (legal basis: contract, 6(1)(b); safety and abuse prevention: legitimate interests, 6(1)(f)).

• Sign-in emails from auth@fluxstudy.com, only when you ask to sign in. They contain a link or code that expires after 5 minutes.
• Study reminders and a weekly recap from notifications@fluxstudy.com, sent outside your quiet hours. Turn them off any time in Settings.
• We don’t send marketing emails, and we don’t track email opens or clicks.

• All connections use HTTPS. Session cookies are HttpOnly, Secure and SameSite.
• Sign-in links and codes are stored hashed, expire after 5 minutes, work once, and are rate-limited.
• Sign-in emails are protected by a human check (Cloudflare Turnstile) against automated abuse.
• Every request is checked against your account: your files, passages and search results are only reachable by you. Stored files are private and not publicly listed.
• Payment card details are handled by Stripe and never touch our servers.

No system is perfectly secure. If a personal data breach is likely to put your rights at risk, we will notify the supervisory authority and, where required, you.

FluxStudy is for adults aged 18 and over. We don’t knowingly collect personal data from anyone under 18. If you believe someone under 18 has an account, contact us and we will delete it.

We use strictly necessary cookies and on-device storage to run FluxStudy and remember settings you change. Optional preference and analytics cookies are only used after you allow them in our cookie banner, and are off by default. We honour Global Privacy Control as a refusal of analytics. There are no advertising cookies.

Analytics will be provided by AppHelm (product analytics), which is not yet enabled. You can change or withdraw your choice at any time through “Cookie settings” in the footer or Settings → Privacy & data; when you withdraw, analytics stops and the analytics cookies we can identify are deleted. See the Cookie & Storage Policy for every item.

If you live in California, our Notice for California Residents explains your rights under the CCPA/CPRA. We do not sell or share personal information.

We’ll update this policy when our practices or providers change, and revise the “last updated” date. If a change materially affects how we use your data, we’ll tell you in the app or by email before it takes effect.