UvA AI Chat

UvA AI Chat offers a safer alternative to commercial models, providing stronger protection for the data and privacy of UvA students and staff. The tool functions in much the same way and is just as user-friendly as commercial models such as ChatGPT, Gemini or Claude.

Privacy

All input from students, lecturers and staff remains within the UvA environment. There is no permanent data sharing with commercial or foreign parties, and no one within the UvA monitors what you enter. User prompts are processed in anonymised form for analysis, reporting or improvement purposes.

Control over data

The UvA considers digital autonomy important, which is why we determine how data are stored, processed and protected. Although we do not control how the underlying language model has been trained, we do control key components such as filtering, permitted queries and system prompts. This enables us to make UvA AI Chat digitally autonomous in the future and, where appropriate, connect it to ethically responsible models such as GPT-NL or EuroLLM.

Accessible

UvA AI Chat has been designed with ease of use in mind within the UvA context. The system can be tailored to specific needs, for example to support the development of teaching materials.

Free access

UvA AI Chat does not operate with different subscription tiers and is available free of charge with a UvANetID. This ensures that everyone has access to the same functionalities and that we do not contribute to information inequality among our students, lecturers and staff.

Academic freedom

By managing the AI system’s front end ourselves, the UvA ensures that the use of UvA AI Chat leaves no room for censorship or external interference.