You don't nuke the accounts, you make it so no one can see their posts and they don't realize. Make them waste their time instead of creating a new account.
What’s the false positive rate, and is that why people complain they’re being banned or shadow banned?
(Definitions of shadow ban vary: https://blog.twitter.com/en_us/topics/company/2018/Setting-t...)
It's been tweeted by multiple security people at various Twitter engineers already and raised in lots of ways. They know and don't care. Meantime, real people lose their money - otherwise nobody would waste their time to post the scam.
You offer users a list of filters they can choose to activate, so they only see content according to the filters they enabled.
There would be filters against spam, filters against opinions other than your own, etc.