I believe there are no exceptions to this rule. That's what I have always read, and I have never encountered any, neither in Greek nor in Latin, nor even in German.
There is an hypothesis about the cause of this phenomenon. Neuter words were historically limited to inanimate objects or things that cannot act. In a basic sentence, it was rarely or never the neuter thing that acted, and therefore you didn't need case to distinguish between agent and patient/theme (between "acting" and "being acted upon"). So, if you just have a transitive verb and a neuter word, there is only one possibility: the neuter thing is acted upon. With an intransitive verb, there is no object, so it is even less necessary. So semantics make case redundant for neuter words.
This hypothesis is inspired by our knowledge of ergative–absolutive languages, in which the 'subject' of an 'intransitive' verb and the 'object' of a 'transitive' verb take the same case (the absolutive case), as opposed to the 'subject' of a 'transitive' verb (ergative case, from ergon "work, deed, action", related to work). In other words, the primary complement of a verb is the thing that does not act; and only some verbs have a secondary complement, which is a thing that acts.
(This contrasts with nominative–accusative languages, such as the main European languages, in which the subject of an intransitive verb takes the nominative, but the object of a transitive verb takes the accusative case. Our primary complement is the subject, which often acts; and only some verbs have a secondary complement, the direct object.)
Of course this is an hypothesis, so take it with a grain of salt. But it's the only one I've ever heard.