There's a flaw in this classic "arbiter of truth" argument. We shouldn't ask "who" can we trust as the arbiter of truth - the answer is we cannot trust anyone with that responsibility. Instead we should ask "what" can we trust?

It's an epistemic question and we have already developed good ways of dealing with it as a species. Look at the scientific method which fundamentally relies on the principle of falsifiability to seek truth. Free speech would ideally rely on the same principle if we consider it useful for the purpose of leading us toward truth. If the speech in question is unfalsifiable or is falsified then it shouldn't be considered as protected speech. This doesn't necessarily mean that it's automatically censored, but that if this speech is challenged for censorship then free speech alone would not be an adequate defence against censorship. There may be other defences for it - for example, we still publish scientific papers which falsify their own hypothesis but we wouldn't keep publishing the same falsified hypothesis repeatedly unless there is new information to bring it back into question in a way which is falsifiable. It's pretty simple and it has already been proven to operate exceedingly well as a method of truth arbitration as demonstrated by its utility with the rapid acceleration of modern science.