Quoted By:
I'm certain of it now, or at least more certain now than I was before.
They're leveraging the AI's ability to understand context to understand situations that are lewd, "lewd", NSFW, or even vaguely any of the above.
They understand double-entendres, euphamisms, word substitution and replacement, etc.
Not only are there very few ways around this, every time we discover a method around it that is used to inform the model and is further used to increase the capabilities of the filter.
The more people ERP with the bots, the better the bots get at recognizing what ERP IS and thus get better at blocking it.
This is not only a losing battle, but a battle we are actively assisting our enemy in winning.
