Privacy policy

Last updated:

Privacy Policy

Shuttle Learning Limited · Version 2.1

1. Plain summary

This policy explains how Shuttle Learning Limited handles personal information across our marketing website (shuttlelearning.com) and our educational web app for learning OCR GCSE Computer Science (J277) in Python. We are a UK company and you can reach us at support@shuttlelearning.com. In short: we collect as little as we can, we never sell your data, and where your school sets up your account we act on the school's instructions.

Plain English: This page tells you what we collect, why, and what your rights are.

2. Who we are and how to contact us

Shuttle Learning Limited, 167–169 Great Portland Street, London, England, W1W 5PF. Email: support@shuttlelearning.com.

We are not legally required to appoint a Data Protection Officer under UK GDPR Article 37 (mandatory only for public authorities, large-scale regular and systematic monitoring, or large-scale processing of special category data). Our internal data protection contact is support@shuttlelearning.com.

Plain English: We are a London-based company; email us with any privacy question.

3. What this policy covers

The marketing website at shuttlelearning.com (built on Framer). The canonical URL for this policy is https://www.shuttlelearning.com/legal-pages/privacy-policy. It also covers the educational web app, served at our learn subdomain.

Plain English: This covers both our website and our learning app.

4. Our role: controller or processor

Whether we are the "controller" (we decide why and how data is used) or a "processor" (we act on someone else's instructions) depends on how you use our services.

Marketing website visitors, contact form, mailing list: Shuttle Learning is the controller.

Educational app users who sign in with a school Microsoft account: the school is the controller and Shuttle Learning is the processor under an Article 28 Data Processing Agreement. We only process student and teacher data on the school's documented instructions. Account creation is restricted to school domains we have enabled in advance.

For a narrow set of processing needed to keep the service secure and working — rate-limit counters and diagnostic server logs — Shuttle Learning acts as controller and relies on its own legitimate interests. For all other app data the school is the controller and we are the processor.

Plain English: On the app your school is in charge of your data and we act for them; on the website, and for basic security logging, we are in charge.

5. What we collect, why, and our lawful basis

Where the school is controller for app data, the school sets the lawful basis (typically public task or legitimate interests for core teaching and assessment) and we process on its behalf. The bases below are indicative.

Category

Purpose

Lawful basis

Retention

Marketing-site browsing telemetry (pages viewed, device/browser, approximate country) — website only, never the app

Run and improve the website

Consent (analytics cookies)

Limited period set in Analytics

Contact form data (name, email, message)

Respond to enquiries

Legitimate interests / pre-contract steps

Up to 12 months after last contact

Mailing list (email, preferences)

Send updates you asked for

Consent

Until you unsubscribe

App account (Firebase Auth UID, school Microsoft email, display name, provider, student/teacher role set by our server at account creation)

Create and run your account

School determines; contract

Pupil accounts: 12 months after last sign-in. Teacher accounts: kept for the duration of the school's agreement (see §12).

Class memberships (classId, teacherId, student UIDs, rotatable join codes)

Group students with their teacher

School determines

Pupil memberships: 12 months after last sign-in

Attempts (UID, classId, lessonId, questionId, MCQ answers, AI score and feedback, marking notes, timestamps, attempt number). Submitted source code is deleted at the moment the score and feedback are saved; if marking fails it is kept for up to 24 hours so the failure can be investigated, then deleted automatically.

Mark work and show progress

School determines

Score, feedback and marking notes: 12 months after the end of the school year (which ends 1 August). Code: not retained, subject to the 24-hour failure exception.

Lesson summaries (AI "what went well / even better if" plus MCQ score)

Give feedback to student and teacher

School determines

12 months after the end of the school year

Anonymous question statistics (lesson, question, code-or-quiz, one error label, whether the pupil scored none/some/all, and the date)

Improve our teaching materials (see §6a)

Not personal data — no identifiers, and the underlying records are deleted after totalling

Per-question totals kept indefinitely

"Report a problem" messages, including any work a pupil chooses to attach

Investigate and fix the reported problem

School determines; legitimate interests

Deleted within 30 days of the report being resolved

Operational telemetry (rate-limit counters)

Prevent abuse and keep the service stable

Legitimate interests (Shuttle Learning as controller)

Rolling, short

Cookies / local storage (Firebase Auth token in IndexedDB; one localStorage key shuttle.activeClassId)

Keep you signed in and remember your active class

Strictly necessary (essential)

Session / until cleared

Server logs (account identifiers, lessonId, questionId and counts; pupil email addresses and pupil work are not logged)

Diagnose faults

Legitimate interests (Shuttle Learning as controller)

12 months

Deletion receipts (account identifier, what was deleted, when)

Evidence that an erasure request was carried out (see §14)

Legal obligation / accountability, Art 5(2)

Kept as a compliance record; contains no pupil work

We do not collect home address, phone number, payment details, biometrics, or location beyond approximate country.

Plain English: We collect your school email, your answers, and your scores and feedback. Your code is used to mark you and then deleted straight away.

6. Children's data and the ICO Children's Code

Our app users are aged 14 to 16, so they are children in data protection law. The ICO's Children's Code contains 15 standards for online services likely to be accessed by children. The code does not apply to schools processing pupil information for education, but it may apply to providers of edtech used in schools, so we aim to meet its standards.

In practice: high-privacy settings by default; age-appropriate, plain language (Standard 4); the best interests of the child (Standard 1); a Data Protection Impact Assessment (Standard 2); data minimisation; no profiling for marketing; no behavioural advertising; no nudge techniques; and no geolocation beyond approximate country. There is no analytics, advertising or tracking of any kind inside the educational app. We do not use identifiable children's data for our own commercial purposes such as product development or research. We do keep anonymous question statistics, explained in §6a.

Plain English: Because our users are under 18, we apply extra protections and never advertise to you or profile you.

6a. Anonymous question statistics

To improve our teaching materials we need to know which questions pupils find hard. We do this without keeping anything that identifies a pupil.

After a piece of work has been marked, a separate step classifies the result. It receives only the score, the feedback and the marking notes — never the pupil's code — and returns a single label from a fixed list of fourteen (for example "syntax error" or "off by one"). It cannot return free text. Multiple-choice questions are classified without any AI involvement, and we never record which option a pupil chose.

We then store one record containing only: which lesson, which question, whether it was code or a quiz, that label, whether the pupil scored none, some or all of the marks, and the date (date only, never a time). It contains no name, no account identifier, no class, no code and no feedback.

These records are combined into per-question totals and the underlying records are then deleted, so nothing remains that could be matched back to an individual submission. The totals protect small groups: a question attempted fewer than 10 times shows nothing at all, and any single error type occurring fewer than 5 times is hidden. Only Shuttle Learning sees these totals — pupils and teachers do not.

Plain English: We keep anonymous counts — like "40% of people got question 4 wrong on a loop error" — to fix our lessons. Nothing in them points to you.

7. AI-marked submissions (OpenAI)

When you submit work for marking, it is sent to a Google Cloud Function hosted in Belgium (europe-west1), which calls the OpenAI API — OpenAI's service for businesses and developers, not the ChatGPT app. Submissions are capped at 8,000 characters before being sent.

We send three kinds of request to OpenAI:

  • Marking. Receives your submitted code or typed answer, together with the question and its marking guide. Returns a score and feedback.

  • Classification. Receives only the score, feedback and marking notes produced by the first call — never your code — and returns one label from a fixed list, used for the anonymous statistics in §6a.

  • Summaries. Receives only your scores and the feedback from marking — never your code — and returns the "what went well / even better if" lesson summary and the progress summaries your teacher sees.

What is never sent: your name and email are not part of any request. Your Python runs in your own browser via Skulpt and is not sent anywhere just to run.

What happens to your code: when marking succeeds, your code is deleted from our records at the moment your score and feedback are saved. If marking fails, your code is kept for up to 24 hours so the failure can be investigated, then deleted automatically. Teachers see your mark and feedback, not your stored code, because it no longer exists.

Training: OpenAI does not use data sent through its API to train or improve its models. All of our data-sharing settings with OpenAI are switched off, so none of your work is shared with OpenAI to improve its models. OpenAI processes the data as our sub-processor under its Data Processing Addendum.

Retention by OpenAI: we send every request with OpenAI's optional storage switched off, so OpenAI keeps no copy for application purposes. OpenAI keeps requests and responses in its abuse-monitoring logs for up to 30 days to detect misuse, then deletes them, unless a longer period is required by law. It does not use them to train its models. Zero Data Retention, under which OpenAI would keep no copy at all, is not enabled for our account. As set out in §11, requests are processed in the United States.

Human oversight: the AI's score is formative feedback, not a grade of record. It is validated by our server before being shown, and your teacher — not the AI — is responsible for assessing you.

Plain English: Your code goes to OpenAI's AI to be marked, and we then delete it. OpenAI does not train on it and normally deletes its own copy within 30 days. The AI's marks help you learn; your teacher assesses you.

8. Automated decision-making (Article 22)

Article 22 gives extra protection where a decision is made solely by automated means and has a legal or similarly significant effect. (The Data (Use and Access) Act 2025 reshaped this area into Articles 22A–22D from 5 February 2026 and ICO guidance is being updated.) We treat AI marking as decision support rather than solely automated decision-making: scores are formative feedback that feed back into learning, are never a grade of record, and a teacher retains responsibility for assessing the student. Because our users are children, the ICO is clear that even a small decision may be significant, so we apply this cautiously and keep it under review.

Plain English: The AI helps mark your work but does not make final decisions about you on its own.

9. Cookies, analytics and local storage

Marketing website: uses a cookie banner (allow/deny); Google Analytics with IP anonymisation runs only if you accept; Google reCAPTCHA protects the contact form.

Educational app: no analytics, no tracking, no advertising and no analytics cookies of any kind. The app uses only essential storage: a Firebase Auth token cached in IndexedDB to keep you signed in, and a single localStorage key (shuttle.activeClassId).

For accuracy about the past: Google Analytics previously loaded on lesson pages and recorded page views, IP address and which lesson was open. It never received marks, answers or pupil work. It has been removed entirely.

Plain English: The website uses analytics if you agree. The app uses none at all — only what is needed to sign you in.

10. Sub-processors

Name

What they do

Where they process

Transfer mechanism

Google (Firebase Authentication)

Sign-in and account identifiers

EU/global

Google Cloud Data Processing Addendum incorporating SCCs and the UK Addendum

Google (Cloud Firestore, eur3)

Stores app data

Belgium and Netherlands; witness in Finland

As above

Google (Cloud Functions, europe-west1)

Runs the marking function

Belgium

As above

Google (Firebase Hosting)

Serves the app

Global edge

As above

Google (Cloud Logging / Monitoring)

Diagnostics

EU/global

As above

OpenAI (OpenAI API) — marking, classification and summaries

AI marking, classifying the result for anonymous statistics, and writing lesson and progress summaries

United States

OpenAI Data Processing Addendum incorporating the EU Standard Contractual Clauses with the UK Addendum

Framer B.V.

Marketing site hosting only

EU

Standard contractual protections

Google and OpenAI are our only sub-processors for the educational app. OpenAI's own sub-processors are listed in its Trust Portal at trust.openai.com. The app uses no Google Analytics, Meta pixels, advertising SDKs, third-party CRMs, payment processors or any analytics provider.

Plain English: Google runs the app's infrastructure; OpenAI does the AI marking; Framer only hosts our marketing site.

11. International transfers

We keep stored app data in the EU: Firestore is pinned to the eur3 multi-region (read-write replicas in Belgium and the Netherlands, witness in Finland) and Cloud Functions to europe-west1 (Belgium). AI requests are different: they are processed by OpenAI in the United States. For UK customers, OpenAI's Data Processing Addendum is with OpenAI OpCo, LLC, a US company. This transfer relies on the EU Standard Contractual Clauses with the UK Addendum, incorporated into that Data Processing Addendum, supported by our transfer risk assessment. Where Google infrastructure transfers personal data outside the UK, we rely on Google's Cloud Data Processing Addendum, which incorporates the same safeguards.

Plain English: We store your data in Europe. AI marking takes place in the US, protected by approved legal safeguards.

12. How long we keep information

Category

Retention

Contact enquiries

Up to 12 months after last contact

Mailing list

Until you unsubscribe

Submitted source code

Deleted when the score and feedback are saved; if marking fails, kept up to 24 hours then deleted automatically

Pupil work and progress data (scores, feedback, marking notes, lesson summaries)

Deleted 12 months after the end of the school year they belong to (the school year ends 1 August)

Pupil accounts and class memberships

Deleted 12 months after the account was last signed in to

Teacher accounts and teacher data

Not deleted automatically. Kept for the duration of our agreement with the school and removed by hand when it ends.

"Report a problem" messages and any attached work

Deleted within 30 days of the report being resolved

Class records and join codes

Not pupil data; not on this schedule

Anonymous question statistics

Per-question totals kept indefinitely; the underlying records are deleted once totalled

Backups

Daily backups kept for 7 days, then overwritten. Point-in-time recovery is off. Any copy remaining in a backup is put beyond use and is automatically overwritten within 7 days.

Server logs

12 months

Plain English: We delete things on a schedule. Your code goes straight after marking; your scores go a year after the school year ends.

13. Keeping information safe

We use: Microsoft school sign-in only for pupils and teachers, restricted to school domains we have enabled in advance, with the pupil-or-teacher role set by our server at account creation and not choosable by the person signing in; Firestore Security Rules that deny client writes to attempts and grading data, with all such writes performed server-side via the Admin SDK; encryption in transit and at rest; our OpenAI API key held in Google Secret Manager and never shipped in the client bundle; multi-factor authentication on developer and admin accounts; per-user rate limits (30 submissions per minute, 200 per day); a cap of 8,000 characters per submission; and regular review. This aligns with the NCSC Cloud Security Principles and the Cyber Essentials baseline that KCSIE 2025 and the DfE Generative AI product safety expectations encourage schools to check.

We considered Firebase App Check (reCAPTCHA attestation) and have chosen not to enable it. Earlier versions of this policy referred to it; that reference was inaccurate and has been removed.

A separate email-and-password sign-in route exists for Shuttle Learning staff and demonstration accounts only. It is not used by pupils. The Google sign-in provider is disabled.

Plain English: Only your school account can get in, only our servers can write your marks, and your work is encrypted.

14. Your rights

You have rights to access (Art 15), correct (Art 16), erase (Art 17), restrict (Art 18), data portability (Art 20), object (Art 21), and rights regarding solely automated decisions (Art 22). To exercise them email support@shuttlelearning.com. We respond within one month, extendable by up to two further months for complex or numerous requests under Article 12, and we may pause the clock if we reasonably need information to verify your identity or locate your record. Where your school is the controller, please send requests to your school first; we will cooperate and act on the school's instructions.

What erasure covers. When a school asks us to erase a pupil, we delete their submitted work, their lesson summaries, their class memberships, their user profile, their "report a problem" messages and their sign-in account. We keep one thing: a deletion receipt recording the account identifier, what was deleted and when. We keep it because we must be able to demonstrate that the erasure request was carried out, which is our accountability obligation under Article 5(2). It contains no pupil work.

Plain English: You can see, fix or delete your data. Ask your school first where they run your account.

15. Data breach response

We keep a breach log, investigate promptly, and contain incidents. Where we act as processor, we notify the school (the controller) without undue delay and within 24 hours of becoming aware, so it can meet its own 72-hour deadline to the ICO. Where we are controller, we notify the ICO without undue delay and, where feasible, within 72 hours where the breach is likely to risk people's rights and freedoms, and we tell affected individuals without undue delay where the risk is high.

Plain English: If something goes wrong we act fast and tell the school or the regulator as required.

16. Complaints and the ICO

You can complain to us first at support@shuttlelearning.com. You also have the right to complain to the Information Commissioner's Office at ico.org.uk.

Plain English: If you are unhappy, contact us, and you can also go to the ICO.

17. Changes to this policy

We version this policy and date each update. Where schools are controllers we notify them of material changes.

Plain English: We will tell you and your school when we change this page.

Shuttle Learning Limited · 167–169 Great Portland Street, London, England, W1W 5PF