Where do you think they get the data from? The entire web of internet has been consumed, these models need constant new data or else they fall behind.
I don't see in anyway they can't not steal your work. It may not be all of it out right, and most peoples projects are not even that important to steal. They might just steal specific functions or copy off some of your work but essentially reskin the rest of it.
But Microsoft's a monopoly and has done this stuff in the past with word, and tried to do it again with the Microsoft store around windows 7/8 era. More and more of their product is being agentically coded, and if a new feature becomes popular enough in third parties they'll likely just spin off their own version. And if you use their services then they just outright have access to your work.
These tech companies like meta just consume anything below it and either kill it or encorperate it into themselves. I don't think its that far of a stretch to think they don't already do this to some extent.