This is the corporate equivalent of "think of the children". A justification that could have been used to gate-keep any and all aspects of computer science, and one that isn't even logically consistent since they only hide their code and weights while still publishing their research: making it ultimately reproducible by malicious actors, especially those well-funded, while slowing down researchers and competitors.
We are privileged to work in a field where we have open access journals, and where there is a large undergoing drive to improve the reproducibility of papers by releasing the code and weights. Their behaviour is the antithesis of what the field is working towards and having talked to many researchers, I don't know many that are fooled by it.