Pretty interesting that everyone here is overwhelmingly against Meta for trying to keep their models private, but the other company (the one with "open" in their name) is applauded for doing exactly this.
While true, I also believe OpenAI at least has a chance of their model being proprietary. So, while I am angered by OpenAI, I am unsure I could make the same to-me-slamdunk argument for why they have no standing for a DMCA on their weights (were such leaked)... whereas, for Facebook, either the weights can't be copywritten at all, they can be copywritten but the copyright is owned by the people who made the training data, or--and this is key--the model is a derivative work of the code they released under GPL.
