A safe learning environment
AI-School is designed specifically for schools, with the option to sign in through the school's environment.
Privacy and responsible use
Using AI safely at school
AI-School helps schools use AI responsibly, with control over access, supervision and data processing. We have agreements with AI model providers on how they process personal data. This ensures that chatting within the school environment is handled with care.
The school board can also decide how AI-School handles personal data in chats: it can be passed on, blocked, replaced or held back. Account details are not shared with the AI provider, but information users enter themselves may be. So only enter what’s needed, and follow your school’s guidelines for audio recordings.
Watch the video to see how it works. Below, you can read more about the measures and agreements.
Download the guide as a PDF
Prefer to read it at your own pace? This PDF gives you a clear overview of how to use AI-School safely.
This PDF is in Dutch.
Safe use at school
Guidance and monitoring
Students receive guidance and are monitored while using the application.
Personal data protection
AI-School uses technical and organisational measures to protect personal data.
Anonymous use is possible
Users can also work with anonymous accounts in the application, without using personal data.
Privacy and data processing
General Data Protection Regulation (GDPR)
What is the GDPR?
Educational institutions collect and use a lot of personal data about their pupils or students. The GDPR requires you to handle this data carefully and protect students' privacy.
How does AI-School support GDPR compliance?
AI-School uses a data processing agreement, roles, retention periods and technical privacy protections. The school remains responsible as the data controller for its own configuration and use.
Which data processing agreement?
AI-School uses the model agreement from Privacyconvenant Onderwijs as the basis for its data processing agreement.
Privacyconvenant OnderwijsWhat happens to submitted chats?
The selected AI model receives the prompt content, relevant chat context and any selected documents or tool results to generate a response. This content is sent without being linked to the AI-School account.
Account and administration data
AI-School processes information such as names, email addresses and student numbers for accounts and administration. This data is not automatically added to the prompt sent to an AI model.
Personal data in a chat or recording?
If a user includes personal data in a prompt, document or tool instruction, the selected model provider may process that data. So don't enter personal data unless it's necessary. Where possible, administrators can enable PII protection. AI-School can't automatically filter personal data from audio files, so follow your school's guidelines for recordings.
Data Protection Officer
Grip IT Consultancy
Patrick Lankhaar, FG0016243. For privacy questions, you can contact the data protection officer directly.
Supervision and guidance
Supervision and content filters
Chat history
Depending on the roles and permissions configured, authorised teachers can view chat history to provide guidance and monitor use.
Live insight during lessons
Teachers can start a lesson and monitor students during the lesson.
Content filters
AI-School and model providers use safety measures and content filters to limit unwanted or risky output. Filters reduce risks, but they can’t prevent every unwanted response.
Sign in with your school account
We connect AI-School to Microsoft or Google so you can securely sign in with your school account. Your school must approve the use of our Microsoft application, or it can submit its own Microsoft application to connect.
The connection checks whether the email address is registered in the AI-School application. This prevents unregistered email addresses from accessing the school environment.
AI-School subscription
View our AI-School subscription terms and conditions, service terms and privacy statement. These documents form part of the agreement with AI-School. We also enter into a data processing agreement with the school board.
Behind the scenes
What happens when I ask a question in the chat?
- 1
Retrieve question
When you ask a new question, the chat’s questions and answers are first retrieved from our database.
- 2
Send securely
The chat history and your new question are encrypted and sent to a language model through an API connection.
- 3
Show answer
The language model generates a response. AI-School receives it and displays it in the chat window.
Where is this information sent?
AI-School offers various language models, such as GPT, Claude and Gemini. The information is sent to the provider that offers the connection for the selected language model. In most cases, this is OpenAI or Google.
What does the language model do?
The language model processes the information and generates a response. These models are complex, and AI-School has no control over how they work or what they produce.
What happens if I turn on image mode?
Image mode generates an image from a description. In image mode, we don’t retrieve chat history. Instead, we send your request directly to the text-to-image model through an API connection.
What is OpenAI?
OpenAI is the company behind ChatGPT. AI-School uses language and image models through API connections to generate responses and images within your school’s own environment.
What data does the model provider receive?
The provider receives the prompt and the context needed to process it, but not your name, email address, student ID or account connection automatically. However, any personal data a user includes in their input may be processed by the provider.
Is model input used for training?
AI-School does not make API input available for model training. For standard business API services, providers don’t use input to train models without explicit permission. Retention periods vary by provider, model and feature, and may include temporary storage to detect misuse.
AI in the classroom
Frequently asked questions
Can children use AI?
Schools must ensure that AI is used appropriately, in line with applicable laws and their own policies. AI-School provides administration, supervision, data processing agreements and technical privacy measures to support this.
What about the European AI Act?
The European AI Act sets rules for AI systems based on their level of risk. AI-School keeps up with these developments and designs its environment to support safe, transparent use with human oversight.
Why not use a personal ChatGPT subscription?
Personal accounts aren’t part of a trusted school environment. AI-School lets schools manage accounts, enables supervision and avoids the need for student data to create external accounts.
Why is supervision needed?
Without supervision, students may misuse AI tools. AI-School provides access to chat history and a live view during lessons, so students can receive appropriate guidance.
Integrations and sub-processors
List of sub-processors
AI-School works with sub-processors for hosting, sign-in and AI models. Below, you’ll find details of what each provider is used for. The locations are based on the Edu-V processor appendix dated 6 July 2026.
OpenAI
We use OpenAI for GPT models. OpenAI processes input, prompts and chat history to generate responses. Processing takes place in Ireland and the United States. We have a data processing agreement with OpenAI.
Google Cloud
We use Google Cloud to host AI-School’s virtual environment. Google Cloud processes names, student IDs, email addresses and chat history. Data is stored and processed in the Netherlands and Belgium.
Google Identity Toolkit
We use Google Identity Toolkit to authenticate users. This involves processing email addresses, IP addresses, user agents and user IDs. Data is stored and processed in the Netherlands and Belgium.
Google Gemini
We use Google Gemini as an AI model that lets users ask questions, generate text and analyse data. Gemini processes input, prompts and chat history. Data is stored in the Netherlands and Belgium; prompts may be processed in the Netherlands, Belgium and the United States.
Anthropic
We use Anthropic for Claude models. Anthropic processes input, prompts and chat history to generate responses. Processing takes place in the United States. We have a Data Processing Addendum with Anthropic. Standard contractual clauses are used for transfers outside the European Economic Area.
Mistral AI
We use Mistral AI for Mistral models. Mistral AI processes input, prompts and chat history to generate responses. Processing takes place in France.
Security measures
How do you secure your systems?
We combine infrastructure security, roles, environment permissions and technical controls to ensure that only users with the right permissions can access data.
Cloud security
We use Google Cloud Platform and Firebase for storage and backend services. Google Cloud follows various standards and certifications, including ISO/IEC 27001, ISO/IEC 27017, ISO/IEC 27018, SOC 1, SOC 2, SOC 3 and the GDPR.
Preventing unauthorised access
Users sign in through Microsoft or Google, are assigned roles such as student, employee, teacher, admin or super admin, and can only access the environment they belong to.
Security Rules and App Check
Firestore Security Rules enforce roles and environment permissions. Firebase App Check helps protect backend resources by checking that requests come from authorised front-end applications.
More about security
Assessments and checks
These assessments describe the situation within their own scope and assessment period. They do not constitute a general certification of AI-School.
DPIA
- Assessor
- Grip IT Consultancy
- Date
- 8 September 2026
- Scope
- Risks associated with collecting data in AI-School: which data is available, who has access, and how data is supplied and retrieved.
Within the data processing assessed, the average privacy risk was rated low and no high residual risks were identified. Schools can request the report.
Vulnerability scan
- Assessor
- Grip IT Consultancy
- Date
- 16 April to 16 June 2025
- Scope
- Remote, partly automated assessment of ai-school-pro.web.app.
The overall score was 8 (low risk), with some points for attention and recommendations. This does not guarantee the absence of vulnerabilities.
Information security risk analysis
- Assessor
- Grip IT Consultancy
- Date
- 19 June 2025
- Scope
- Threats, existing security measures and residual risks of AI-School.
The analysis identifies, among other things, the risk of a user including personal data in a prompt. The findings relate to the measures assessed.
Still have questions about privacy?
Email or call our data protection officer at Grip IT directly.