New AI platform 'Nebula' launched within VU for safe use in research and innovation
Radu Apșan, Scientific Software Engineer for Nebula
Nebula is a new AI platform that enables researchers to work with advanced language models (LLMs) without completely relying on commercial providers. Especially valuable for interdisciplinary collaborations and projects beyond the university, Nebula ensures control over data, scientific standards, and research integrity. Who else can truly safeguard these values?
Data means real people
Imagine this: a crucial research project at the intersection of criminology and judicial practice requires analyzing a database of actual prisoners. Real people, real lives. In this scenario, using a commercial AI platform is simply out of the question. Control over data is therefore essential, yet in the brave new world of AI, researchers should not have to miss out on new possibilities.
Key issues: firm commercial positions and blackbox models
Most AI models are hosted by large tech companies outside Europe, placing control over data, privacy, and ethical standards firmly in their hands instead of that of the users’. Whether it’s GPT, Claude, or Gemini, these are proprietary, black-box systems. And that matters: when the inner workings of a model aren’t transparent, core scientific principles like replicability can’t be guaranteed.
Against this backdrop, Nebula was developed at Vrije Universiteit Amsterdam by the Network Institute in response to a growing need amongst researchers.
A first in the Netherlands—and likely across the EU
Unlike proprietary, black-box systems, Nebula is entirely open and locally hosted, ensuring full control over data, compliance with EU regulations, and scientific replicability. With over 1,500 open AI models available—six already running, including Deepseek, LLaMA, and Qwen—Nebula demonstrates the strength of open-source AI.
Generative AI is already at the core of many scientific fields nowadays, Nebula enables VU researchers to do it in a fully replicable, fair, and ethical way.
