(Thanks to M.C. for showing me the article.)
So in a 2007 article in Analysis*, Peter Milne suggests that omniscient beings would have to be dialetheists, given sentence S, below:
Sentence S: No omniscient being knows that which sentence S expresses.
His reasoning is that, if no omniscient beings exist, sentence S is "quite straightforwardly true." If an omniscient being exists, however, that being would have to both know and not know that which sentence S expresses. Thus, Milne seems to be suggesting, we have to choose between atheism and dialetheism.
Now, I am, to a slightly obnoxious degree, convinced of the deep irrationality of theism, but this particular argument doesn't strike me as a good one. Here's why: if this should force theists to accept true contradictions, then very mundanely familiar parallel arguments should force atheists to accept true contradictions as well. Whatever one thinks about the rationality of believing inconsistencies, the theist doesn't seem to be in any worse a boat than the rest of us.
Why, after all, should we think that an omniscient being would both know and not know the content of S? Because, presumably, if an omniscient being existed and didn't know the content of S, then S would be true, which would mean that omniscient beings would know it.
Why, however, does that last step--that "which would mean"--go through? Presumably because "some being is omniscient" simply means "some being knows the content of all true sentences."
Given this, on the assumption that an omniscient being exists, the claim that no omniscient being knows the content of S reduces to the claim that S isn't true. Put differently, it looks like S is the negation of the conjunction of the claim that there is an omniscient being and the claim that S is true...and is, thus, a pretty straightforward Contingent Liar. Presumably, if Liar sentences in general are problematic in some way--ungrounded, non-truth-evaluable, meaningless, whatever one's story may be--then S is problematic in exactly the same way. The fact that this particular Liar is rigged so as to have its paradoxicality not "kick in" unless an omniscient being exists changes nothing, just as, presumably, any plausible story about what makes Liars problematic should also apply to resolutely non-paradoxical cousins of the Liar like the Truth-Teller sentence ("This sentence is true.")
*"Omniscient Beings Are Dialetheists," Analysis 67.3, July 2007, pp. 250–51.