The problem is, if you are transparent about your constraints, then users who are using your subscription in bad faith and against the terms, they know exactly how to maximize usage.

It's the same thing when people say that Gmail ought to publish the rules they use for blacklisting senders. If they did, then there would be a lot more senders abusing email.

Whenever you are defining rules internally for catching bad actors, you cannot make those rules public. It defeats the entire purpose.

So maybe Anthropic is losing good will, but it's better than the alternatives.