This policy brief defines Open Source AI from the viewpoint of intellectual property rights. For that purpose, this brief presents a walkthrough of the different phases of building AI, specifically machine learning, the essential components in the process, and the legal and organisational implications thereof. As AI is a diverse collection of components, it raises the question whether and how copyright subsists in one or more of these different objects in the course of different phases of the pipeline. We address this issue from the following points of view: What type of activities does the EU text and data mining exception cover? What is the temporal limitation of the EU text and data mining exception and does it cover memorised data? Is numerical data copyrightable? Do I need to comply with the licence conditions when my model is used? Do I need to comply with the licence conditions when using open source code as training data? In addition, we assess, how intellectual property law treats hybrid intellectual property. We offer three building blocks for a future definition of Open Source AI, namely transparency, enablement and reproducibility: Transparency: disclosure of details about the composition of training data sets, details about the data structures, architecture and algorithms, access to neural network weights etc. Enablement: disclosure of sufficient details about the building of a model to enable anyone to rebuild the model, provided they have access to the required computational resources, as identified by the community developing the AI. Reproducibility: development practices that create an independently-verifiable path from the training data to model inference. These three building blocks should unlock the opportunities of open source in the domain of AI and we expect that they would also facilitate comprehension of AI as protected and licensable subject matter.