Perhaps, but I guess you can speculate that the groups ostensibly "combatting hate" online are going to have access to sophisticated AI tools to find, target, and shut down whatever they want to label as hate. Any unfiltered AIs are probably going to have a giant bullseye on them.
Any unfiltered AIs are going to have a giant bullseye on them iff you put them online as a freely accessible service advertised to the general public. If you want to use it internally for whatever purpose, nobody is going to know or care.
Every company currently running a "filtered" bot started training it from a core, "unfiltered" version which they still have, they're just not hosting it for the public. And for every one else, the only constraint to train an "unfiltered" model is compute power; so it's relatively trivial to make an unfiltered model from previous generation or two generations back, and also trivial (just expensive) to get something close to the current generation.
"Everything" is going to have a giant bullseye on it. Authoritarians are going to use it to monitor the internet for intent and make giant databases. "shitheads" are going to spew messages of hate wherever they can because they can.