This UiB AI seminar is organized by the Faculty of Law and raises the question: Can AI models be open in the same way as open-source software, and should they be?

Open artificial intelligence refers to AI systems whose underlying programs and infrastructure, training data, model architecture, and model weights are wholly or partly openly available, allowing anyone to study, modify, and redistribute them. Large models such as ChatGPT, Claude, and Gemini offer access to their models, but this access is limited to using the model. Other actors, such as Meta’s Llama and the Chinese model DeepSeek, present themselves as open, but in reality provide only access to model weights and the ability to run the model locally. Their use is also subject to license terms that restrict both further distribution and commercial use.

But what does it actually mean for an AI model to be open? Can there be limitations on the use of open AI models? And should such models really be open, given that openness may also make misuse easier?

The seminar will be streamed and recorded and given in Norwegian.

PROGRAM:

10:00 Coffee/tea and mingling

10:15–11:30 Presentations

• Torger Kielland, Professor at the Faculty of Law

• Pål Grønås Drange, Associate Professor at the Department of Informatics, Faculty of Science and Technology

11:30 Light lunch and mingling