This is an argument in bad faith but at this point I have zero trust in corporations and feel like you can generally count on them to do shitty things if they can benefit from it so I can be easily swayed by little proof at this point.
If the latter, there are many laws that say you can own an idea, provided it exists somewhere.
But the big news is that it works, just a bit of data can have a large impact on the open source LLMs. OpenAI can't have a moat in their proprietary RLHF dataset. Public models leak, they can be distilled.