There should be some reciprocity here… if you scrape open community content, you should allow the same (or not do it in the first place).

FB is all about the open internet to suit their purposes but yet fiercely guard their walled garden.

How unfair indeed of them to release the model as open source with an accompanying freely available paper explaining all the details of it

The intellectual quality in their walled garden is too low to train a chatbot for "empathy, knowledge and personality".