Kristijan Zimmer

Croatian Academic and Research Network – CARNET

Kristijan Zimmer has been working in CARNET’s Sector for Artificial Intelligence last two years, following 15 years at the University of Zagreb Faculty of Electrical Engineering and Computing (FER) in the Information Support Center, 1.5 years at the Faculty of Organization and Informatics (FOI) on the e-Schools project, and later as an independent senior IT consultant. At CARNET, he focuses on the digital transformation of higher education, the development of digital maturity models, and the integration of artificial intelligence in education processes and tools. He has been the President of the Croatian Information Technology Society since 2016 and HrOpen volunteer in different roles since 1995.

Tokenizing the Myth of Openness: What Makes an LLM Truly Open?

In the age of artificial intelligence, the term “open-source” is increasingly used — and often misused — in reference to large language models (LLMs). This presentation explores the complexities of what it means for an AI model to be truly open, going far beyond publishing and licensing the source code. Presentation dives into the key components of LLMs — including datasets, model weights, training processes, and licensing — and examine why full transparency is rare. Using recent case studies it reveals the strategic, legal, and ethical dilemmas behind so-called “open” AI.

The presentation also introduces frameworks like the Model Openness Framework (MOF) and tools such as IsItOpen.ai to help professionals assess and compare model openness. Through this lens, it questions whether current trends genuinely democratize AI or simply reinforce existing power structures. Finally, attendees will gain hands-on advice on how to experiment with open LLMs on personal devices and explore the evolving open-source AI ecosystem. It is aimed at educators, researchers, developers, and policy-makers interested in transparency, access, and control in the next wave of artificial intelligence.