The automated removal can also come with broad reasoning in response, and that doesn't overwhelm the system.
"Violating Policy X" (where X can be something like Twitter's Rules, not necessarily something specific) is still a terrible explanation, but isn't nothing either. Vague enough spammers can't use it to game the system, specific enough that user's can guess they did something wrong.
If it isn't "specific enough the user can guess what they did", like in your case, then it's a clear failure. So is a lack of a transparent appeal process.
Automation-assisted, not automation everything.
I wasn't advocating for blanket terms that mean nothing. You need to provide the user with something meaningful.
"Violating Policy X" (where X can be something like Twitter's Rules, not necessarily something specific) is still a terrible explanation, but isn't nothing either. Vague enough spammers can't use it to game the system, specific enough that user's can guess they did something wrong.