Yeah, we were always really clear about that, the GitHub is clearly CompVis for example.
We fund dozens of models and collaborate on them too, the ones we assert IP over are those in our GitHub as benchmark models based on open research coded and trained from scratch.
It gets really complex otherwise when you have lots of collaborators eg on the LAION models.
Where is the evidence of this? The article certainly doesn't contain anything that proves they claimed the owned the IP.