Let me briefly sketch the context. Davidson's principle of charity is part of his semantic project, a theory of meaning. It is meant to reconcile semantic compositionality, the idea that meanings of complex expressions are composites of the meanings of their constituents, with the epistemological holism, which Davidson inherited from his teacher Quine, resulting in the indeterminacy of translation. The latter is in tension with the logical atomism, the idea that truth-conditions can be specified for atomic parts of language, from which the rest of it is composed. Nonetheless, a non-compositional language would not be learnable, goes the argument. Compositionality is strongly supported by semantic realists, who also argue for validity of a more or less atomic reference. Holism, on the other hand, is more common in pragmatist semantics since practice presumably validates linguistic behavior as a whole, and not by pieces.
Davidson tried to steer a middle path between the two. His idea was that while the theory of meaning we aim at is compositional (and truth-conditional in the sense of Frege) the way we develop it is holistic. Namely, while we interpret the utterances of "natives" in relation to the (extensionally describable) circumstances surrounding them, and expect inferred meanings to compose, we assign the meanings so as to validate what is true (or at least explicable) by our own lights. In other words, charity serves as a global constraint on translation:
"If we can produce a theory that reconciles charity and the formal conditions for a theory [extensionality and compositionality], we have done all that could be done to ensure communication. Nothing more is possible, and nothing more is needed."
Needless to say, Davidson's semantics came under attack from both sides. Glock's Quine and Davidson on Language, Thought and Reality gives a Wittgensteinian critique of both, focused on their "physicalism", insistence on extensional reduction of intensional terms, and the unrealistic consequences that it entails, including charity.
"Quine’s method of translation cannot yield even the meager results it is supposed to, without tacitly smuggling in either a prior understanding of the natives, or hermeneutical methods and intensional notions which he disowns... the only alternative to taking this kind of understanding for granted is to assume that the native knows that the radical translator is trying to establish the stimulus-meaning of her words. For Quinean translation to work, the natives had better read a translation of Word and Object!
[...Charity suggests that] radical interpretation starts by projecting all of our beliefs on to the natives, allowing for the ascription of error only as a second step... It is glaringly obvious that this is not how we interpret from scratch. The Spanish conquistadors... did not start out on the assumption that the natives shared all of their beliefs, including... that ships can sail against the wind, or that infectious diseases are caused by micro-organisms."
On the other hand, philosophers of "content", semantic byproduct of the computational theory of mind, like Fodor, Dretske and Lepore, advocated a much more referential and representational type of semantics, which includes Mentalese and reduction of intentionality to informational correlations. Lepore and Ludwig's in their book about Davidson question his justifications for the principle of charity, like the idea that identification of beliefs "must depend on a background of largely unmentioned and unquestioned true beliefs", or that speakers must have mostly true beliefs just in virtue of being interpretable as speakers:
"It is only the general beliefs which we must count as true to see you as possessing the concept: we may have to see you as having many related particular beliefs if you have one, but seeing them as true is not a condition on seeing you as possessing the concepts expressed in them... A Cartesian skeptic would find no comfort in an argument based on what is required for concept possession, for he could grant everything it entailed without having been given a reason to suppose correct any of his beliefs about the way things actually were in his environment (if any)."
"[...] We can grant for the sake of argument that if he were in a position to communicate with others, he would have to have mostly correct beliefs about his environment. But why could he not be unfavorably placed, in part because he has mostly false beliefs? Traditional skeptical scenarios employed by philosophers in raising the problem of radical skepticism about the external world will serve as models of how someone could be so unfavorably placed. So, there is no direct route from the characterization of someone as a speaker to the conclusion that he is mostly right about his environment."
To conclude, the demand for the kind of middle ground that Davidson sought to claim is much diminished today. Those partial to scientific realism see little reason to worry about the indeterminacy of translation and holism, discredited by their abuse in postmodernism. The informational semantics is also seen as much better aligned with the current neuroscience. On the other hand, even many naturalized epistemologists view radical translation, and elimination of intensional notions in cognitive sciences, as artifical and unrealistic, see Zammito's Nice Derangement of Epistemes, Ch.2. Also, semantic pragmatists, like Brandom, now confronted the compositionality issue head on, arguing that non-compositional languages can be learned through a sequence of reduxes, each simpler than the next, but all non-compositional, see Brandom's reply to Fodor and Lepore in Reading Brandom.
