I have a specific case here for which I could not find a clear answer here or on other pages. I am working for a company that is interested in using an open source code licensed under GPLv2. The software has machine learning and database components. The machine learning models are used, often in combination with the database, to solve specific tasks. My employer would like to use it internally for its own research. However, we would like to update it so that it works better for us. The modification would aim at improving the models and the database content, so that they are more accurate for the company. This will of course also force us to change the original source code so as to make use of the models. The modified version would be made available on an internal service platform that scientists will use only for internal research. There is no intention of selling/redistributing the software commercially.
Does the company have to release the updates, since technically, the models and the database will be plugged into the source code? If we were to (a) refactor the code to make it flexible so that it would allow any user for dynamically plugging their own sets of machine learning models they train and a modified version of the database, and (b) release the refactored source code under GPLv2: Would we have to release the ML models or any addition to the database content. Your help would be appreciated.