If the most advanced AI models are open source, the existential risk may be higher. Yet, if they are closed, this might lead to too much control from a few organisations or countries.