Nakijken keeps student data on your Mac. This policy explains when the app connects to online services and what must be in place before any cloud feature can be used with student work.
Student data stays on your Mac
Scans, class lists, tests, mark schemes, scores and comments are stored in your project file. Name recognition and matching against the class list always take place on your Mac.
- Qwen and the available Apple Intelligence fallback process recognition and comments on your Mac. The teacher reviews suggestions and decides how to grade the work.
- You choose where to store projects and exports. Saved class and test lists stay in local app storage. If you use a synced folder, share an export or make a backup, the storage provider’s settings and your school’s rules apply.
- When a school uses Nakijken, it determines which student data it needs, who can access it and how long to keep it. Direct questions about student data to the school or its data protection officer.
Signing in is optional
Local features work without an account. Nakijken accounts are available by invitation and are separate from the central account at jessehertogs.eu. If your Mac is still signed in with an older central account, the app checks that session with the central account service.
- The Nakijken account service stores your account ID, email address, optional display name, a hash of your password and the data needed to verify your session. The app stores the session token in the macOS Keychain. A new session in the Mac app is valid for up to one year (365 days) from sign-in. Existing sessions keep their original expiry date. The account website uses a session cookie that is valid for up to 8 hours. Signing out of the app removes the local token and asks the account service to revoke the session.
- Proton Mail delivers invitations, password reset emails and confirmations of a new email address. These messages use one-time links. An invitation is valid for up to seven days, and a password reset or confirmation link for up to one hour.
- You can change your details, check your balance and delete your account on your account page. Deleting your account revokes your sessions and removes your details from the account database. This does not automatically delete the separate quota and cloud usage records, which include the account ID, task status and token usage.
- Session data is necessary to provide the requested sign-in service (GDPR Article 6(1)(b)). The account service may process technical data on the basis of the legitimate interest in keeping sign-in secure (Article 6(1)(f)).
Feedback requires a category, subject and message. You may add your name and email address for a personal reply. A trial account request requires your name and email address.
- Your message or request is sent to the administrator through Proton Mail. Feedback contact details are not filled in from your account. After the delivery attempt, the feedback text and contact fields are removed from the temporary delivery record. Derived receipt and content keys prevent duplicate submissions and are removed from 48 hours onward.
- Unprocessed trial account requests are removed after 90 days. After approval, your name is kept temporarily so it can be attached to your account on activation; that temporary copy is also removed after 90 days. If invitation processing remains interrupted, the request is stopped after 90 days. Its outcome remains available for review for a further 90 days; no invitation is sent automatically.
- Newsletter subscription is optional and is off when an account is created. Only active subscribed accounts receive a newsletter at their current email address. Each newsletter includes a personal unsubscribe link. Only confirmation on the unsubscribe page changes the subscription; opening a link in an email scanner does not. The subscription and unsubscribe key stay with the account and are removed when the account is deleted.
- The content and delivery records of a completed newsletter are removed after 90 days. Uncertain or long-pending deliveries are stopped first; recipient outcomes then remain available for review for a further 90 days. Stopped deliveries never resume automatically. Messages in your own inbox follow your email provider’s settings.
- Forms limit repeated submissions by session, email address and network address. Nakijken keeps derived keys with short expiry periods for this purpose. The protection uses no permanent device identifier and shows no captcha by default.
Updates and additional statistics
- A built-in update check shares the public app version and build with the update service. These requests are counted by UTC day and version as update observations. They are requests, not unique people or installations. You can turn off automatic checks; a manual check still shares the version information.
- The additional opt-in weekly measurement is not currently connected. The server receives no weekly identifiers for it, so the participating-installation charts remain empty.
- Daily update-observation and download-start totals are kept for 13 months. They contain no account, email address, school, project, student data or grades, and are not added together as unique users.
- Website and updater downloads are counted separately when the server begins a full file response. A start proves neither a completed download nor an installation. Resumes, partial downloads, previews and error responses are excluded. Deliveries served by a network cache elsewhere may fall outside this measurement point.
- The retention periods above are deletion thresholds. Scheduled cleanup runs daily and removes records once their threshold is reached. Maintenance or an error postpones cleanup until the next successful run.
- These periods apply to Nakijken’s form and statistics storage. Hosting, network and email providers also process technical data for delivery and security. The periods above do not specify retention for their separate network logs, email archives or backups.
Website, updates and model download
When you visit the website, check for or download an update, or open an online information page, the relevant web service receives technical request data, such as your IP address, the time and the requested file.
- The website and updates use mijn.host for hosting and Cloudflare for network services. Technical data is used for delivery, availability, security and troubleshooting. Retention depends on what those purposes require and on the arrangements with the service provider. The legal basis is the legitimate interest in delivering and maintaining the website and app securely (GDPR Article 6(1)(f)).
- If you choose to download Qwen separately, Nakijken retrieves the model files directly from Hugging Face. Hugging Face receives technical connection data for the download under its own privacy policy. The model then runs on your Mac. The download request contains no student work, account token or project data. Hugging Face privacy policy.
- You can ask the privacy contact below about recipients, retention or safeguards if data is processed outside the European Economic Area.
Restricted cloud pilot
The restricted cloud pilot is for approved users. They can use real student work after checking the selected images for personal data. Names and the mapping to students remain local.
- For a pilot task, the user selects the work locally and confirms the images shown before sending. A request can contain page images or crops of the selected work and the test, as well as the full mark scheme as page images, question numbers, maximum marks and temporary random identifiers. Student names from the class list, the Excel class list, original PDF files, filenames, PDF metadata and the mapping between identifiers and students are excluded. Your email address and session token are never sent to OpenAI.
- When the app recognises a cover page, it excludes the top half on your Mac. The app blocks a request if a possible cover page is uncertain or if the local check detects identifying text. Handwritten names or other identifying information may also appear outside the name field. Cropping and random identifiers therefore do not automatically anonymise images. GDPR, recital 26 · EDPB on anonymisation and pseudonymisation.
- Before real student work can be used with a cloud feature, the school must decide whether the processing is permitted and what data protection arrangements are needed. Technical access does not replace the school's permission and data protection arrangements. A data processing agreement is required where a provider processes personal data on behalf of a controller. Dutch Data Protection Authority on data processing agreements.
- OpenAI processes the selected images through its EU API. Every model request uses store: false, so API responses are not stored as response history. OpenAI may retain content for abuse and safety monitoring for up to 30 days by default, or longer for legal or safety exceptions. This is not Zero Data Retention. API content is not used for model training unless separately opted in. OpenAI data controls.
Contact and your rights
Jesse Hertogs is the contact for processing by Nakijken and its website and account service. For student data managed by a school, contact the school.
- For questions or to exercise your rights over account or website data, email nakijken@jessehertogs.eu. Please keep student work out of your initial message.
- Depending on the processing, you can request access, correction, deletion, restriction or portability of your data. You can object to processing based on legitimate interests. You can also lodge a complaint with your supervisory authority, including the Dutch Data Protection Authority. Lodge a complaint.