Our promise about AI and your data
AI in education calls for honesty about what happens under the hood. Below we explain exactly which AI models LectaMe uses, where your data is processed, and which contractual agreements we have made to keep customer data out of model training. For teachers, schools and businesses that want to know why they can use LectaMe safely.
No training with customer data
Contractually agreed with all AI providers. Your prompts, slides and participant data remain yours.
EU hosting
Persistent data in Amsterdam via Supabase. AI processing with EU providers where possible.
Students without an account
No email address, no tracking, no marketing database. All participant data flows through the teacher.
Encrypted at rest
Disk encryption, TLS in transit, RLS policies at database level. SSL grade A+.
Which AI models we use
Text generation runs through a chain of three providers: Groq (United States), Mistral AI (France) and Cloudflare Workers AI. A request goes to Groq by default; if it hits its limit or fails, the next one takes over. Image generation via several open and commercial models — the choice may change as better options become available. Transcription via Whisper. With every provider, reuse of input for model training is switched off. We publish on /nieuws as soon as there are material model changes.
Where your data is processed
Persistently stored data (presentations, video, images, account details) is held in Amsterdam via Supabase — EU, GDPR-compliant. AI requests go to providers with an EU presence where possible. The full sub-processor list is in the data processing agreement.
No model training with customer data
With all our AI providers we have contractually agreed that customer data is not used for further training or fine-tuning. This applies to prompts, slides, video, images, participant data and analytics. We only send what is needed to handle your request.
Transparency about changes
If we change provider, we publish it in our newsletter and on /nieuws. If we change the purpose of processing, we inform account holders proactively by email. No silent changes via terms updates.
Your rights under the GDPR
Access, rectification, erasure, data portability, restriction of processing — all via /settings or via privacy@lectame.com. For schools, the data processing agreement applies, which formally sets out these rights.
Frequently asked questions
Which AI models does LectaMe use?+
For text generation (slides, cases, quiz questions) LectaMe uses a chain of three providers. Groq by default (model openai/gpt-oss-120b), which processes in the United States under Standard Contractual Clauses; the interface therefore warns against entering personal data. If Groq hits its limit or fails, Mistral AI takes over — a French provider processing within the EEA — and after that Cloudflare Workers AI. For sensitive documents, such as in the overlap checker, and for the learning-goal and growth coaches, the model runs locally on our own server; that content never leaves the machine. For image generation we switch between open and commercial models based on quality; for transcription and subtitling we use Whisper. With all providers, reuse of input for model training is switched off.
Is my data used to train AI models?+
No. LectaMe has contractually agreed with all AI providers that customer data (your prompts, presentations, videos, images and participant data) is not used for model training. We only send what is needed to handle your request, and the provider stores nothing for other purposes.
Where is my data physically stored?+
Persistently stored data (presentations, videos, images, account details) is held in Amsterdam (EU, Germany) via Supabase. AI processing takes place with providers that have an EU presence where possible. The full processing list is in our data processing agreement, which you can download via /verwerkersovereenkomst.
What happens to participant data in live sessions?+
Students and participants do not log in and do not provide personal data. A session code or QR scan is enough. Their answers are stored anonymised under the teacher's session — no email address, no tracking pixel, no marketing database. For minors this is a deliberate design: all data flows through the teacher.
May I use AI-generated lesson content commercially?+
For educational use in your lessons, training and courses: yes, without restriction. For commercial use outside your teaching environment (for example a book or course you sell) the terms depend on the underlying model. The general rule: generated content is yours, provided you use it within LectaMe and in an educational context.
How can I have my data deleted?+
You can delete your account at any time via /settings. Your personal data and your own presentations, media and sessions are then erased; files disappear permanently after 30 days. Work created within an organisation stays with that organisation without your name, as do finalised reports that must remain as evidence. For schools the data processing agreement applies: on contract termination we delete the data within 30 days, sooner on request. Students have no account and therefore no separate opt-out — their data disappears when the teacher deletes the session or presentation.
What if the AI hallucinates or generates factually incorrect lesson content?+
LectaMe's AI is not a knowledge source — it is a didactic structure generator. You as the teacher remain responsible for factual accuracy: review the generated content before you use it in your lesson. For subjects with a high risk of errors (medicine, legal, financial) the rule is: use the output as a skeleton and fill in the facts yourself based on verified sources.
How can I make a responsible disclosure?+
For security reports: security@lectame.com or via /.well-known/security.txt. For data incidents that affect you: privacy@lectame.com. We acknowledge reports within 24 hours and keep you informed about the handling. Reproducible vulnerabilities get a hat-tip in the release notes.
Does the AI assess me, or my staff?+
No. In our mentoring and development environments the AI asks questions and makes suggestions; it does not judge and does not decide. Every suggestion is marked as a suggestion, with its sources, and can be edited or discarded. An assessment is always recorded by a person. Looking back across several moments can only be done by the person concerned, over pieces they select themselves — a manager or assessor cannot use that function. And what is written down as private never goes to an AI model.