the issue is a matter of principles and hypocrisy.

you should not enclose the commons.

that is what openai and anthropic have done; capture the commons, lock the model, restrict the outputs.