Typical forPrivate and boarding schoolsLanguage schoolsTraining and continuing educationDriving schools
AI in teaching. Students are assessed by people.
Give feedback on assignments, answer questions about courses and keep on top of admissions. GoalCube hides students' names and dates of birth before the AI sees anything – and when the AI helps with admissions or grades, a teacher decides.
- Feedback in minutesThe AI suggests feedback; the teacher edits and sends.
- Children and young people protectedNames and ages are hidden. The AI only sees the assignment.
- The teacher decidesAdmissions and grades are high-risk. GoalCube makes sure a person decides.
| Applicant | Born | Test level |
|---|---|---|
| Agnieszka Kowalska | 1994-05-21 | B2 |
| Carlos Fernández | 2001-12-02 | A2 |
| Fatima Rahman | 1988-08-08 | B1 |
Motivation My name is Agnieszka Kowalska and I work as a nurse. I need English for my job and can only come in the evenings.
| Applicant | Born | Test level |
|---|---|---|
| [PERSON_1] | 30–39 years | B2 |
| [PERSON_2] | 20–29 years | A2 |
| [PERSON_3] | 30–39 years | B1 |
Motivation My name is [PERSON_1] and I work as a nurse. I need English for my job and can only come in the evenings.
Agnieszka Kowalska fits the B2 evening class. Carlos Fernández should join A2 with extra pronunciation. Fatima Rahman only wants weekend classes; offer B1 on Saturdays. A teacher decides the placement.
[PERSON_1] fits the B2 evening class. [PERSON_2] should join A2 with extra pronunciation. [PERSON_3] only wants weekend classes; offer B1 on Saturdays. A teacher decides the placement.
Fictitious data. Point at a name or a pseudonym to see what belongs together.
What can you use AI for?
Choose a task. GoalCube works out the risk under the EU AI Act with fixed rules and tells you what to do.
Low risk when the teacher reads the feedback and stands behind it. Names are hidden.
Access to and placement in education are listed in Annex III of the AI Act. A person decides on each applicant.
Evaluating learning outcomes is high-risk. The teacher must decide.
Emotion recognition in education is prohibited (Article 5 AI Act).
Monitoring prohibited behaviour during tests is listed in Annex III.
The example is simplified. In GoalCube you answer a few questions, and the classification follows fixed rules.
What the law says
Education is high-risk
AI for admissions, placement, grading and exam monitoring is listed in Annex III of the AI Act. The requirements apply from 2 December 2027.
Annex III, point 3, AI Act
Emotion recognition is prohibited
AI that reads students' emotions in class has been prohibited since 2 February 2025.
Article 5(1)(f) AI Act
Children deserve extra protection
Data about children needs special care, and information must be easy to understand.
Recital 38 and Article 12 GDPR
Short and simplified. GoalCube helps you follow the rules; it is not legal advice.
Who gets a place in the class? A teacher decides.
The AI has suggested a placement based on test and motivation, without names. The teacher decides on each applicant.
Case facts
Agnieszka Kowalska
- Test level
- B2
- Preferred time
- Evening
- From the application
- Needs English as a nurse.
The AI’s recommendation
B2 class, evening
Reasoning: Test level B2 and can only come in the evenings.
The AI is unsure about:
Audit log
- Decisions are saved here with the time and a timestamp.
Your staff keep working. GoalCube keeps track of the rest.
Low risk is approved automatically under fixed rules. Anything that needs a human lands on This week’s list – typically around 5 minutes a week.
Place the 46 applicants in classes based on test results and preferences.
Everyone has a suggested class. The teacher decides at GoalCube; I haven't seen the names.
4 awaiting a human · approx. 5 minutes
Waiting for you
- Decide class placement for 46 applicantsEducation · the AI has recommended
- Confirm new use: feedback on assignmentsLow risk · approve the rule
Approved automatically
- Summary of customer enquiriesLow risk · approved automatically on Monday
- Overview of supplier contractsLow risk · approved automatically on Tuesday
What you need
Start free with the overview. Add protection when you want to use AI on real data.
AI Map
Map
0 € per month
- AI inventory and AI policy
- Training with proof
- Deadlines, templates and report
Data Vault
Map and protect
from 9 € per month
- Everything in AI Map
- AI on your data, data stays in the EU
- Secret terms and documents
Data Vault + Human Decision
GoalCube for HR
from 15 € per month
- Everything in Data Vault
- A person decides on cases about people
- Passkey, log and timestamps
See all prices All prices are per month and exclude VAT. Pay yearly and get 2 months free. Prices apply from launch, which we expect in November 2026.
Questions and answers
Can teachers use AI for feedback?
Yes. It's low risk when the teacher reads the feedback and stands behind it. Don't put students' names into the AI.
Is using AI for grades high-risk?
Yes. Evaluating learning outcomes is listed in Annex III of the AI Act. The requirements apply from 2 December 2027, and a teacher should decide already now.
Can we use AI to detect cheating?
AI that monitors students during exams is high-risk. Emotion recognition in education is prohibited outright.
Read more
- AI in practice
AI in education: What can schools and training providers use AI for?
2 min readFeedback is low risk, while admissions, grading and exam proctoring are high-risk, and emotion recognition is prohibited. The rules for schools and courses.
- The EU AI Act explained
When is AI high-risk? Annex III explained for ordinary businesses
3 min readRecruitment, HR decisions, credit, insurance and education are high-risk under the AI Act. See the list, the exceptions and what to do from December 2027.
- The EU AI Act explained
Which AI is prohibited under Article 5 of the AI Act?
3 min readEmotion recognition at work, social scoring and manipulation are banned. Here is the full list, with examples from ordinary businesses.
- The EU AI Act explained
Human oversight of AI: what does the law require, and how do you do it?
3 min readHigh-risk AI needs human oversight: a person who understands it, can step in and can say no. What that looks like in practice – and how to document it.
Other solutions for customers
- Online shop customers
- Guests
- Software users
- Members
- Subscribers
- Patients and clients in healthcare
- Clients of advisers
- Tenants
- Policyholders
- Borrowers and credit customers
- Passengers and travellers
- Donors and supporters
- Event attendees and registrations
- B2B customers
- Property buyers and sellers
- Residents in care and supported living
- Citizens
Ready to use AI on student data?
We expect to launch in November 2026. Join the waitlist and we’ll let you know when you can get started.