It's hard to see how placing trust in a giant AI would solve the issue. Just offering "fact-checks" that don't deal in context leaves so much open to interpretation. The major leftist/rightist media networks do what they do not as much through blatant lying (though that happens) as with technically-defensible interpretations of factual events, selective coverage/reporting, and twists of terminology (democracy vs populism, government vs regime, article vs screed, etc). When people call upon services like fact-checkers or find individual studies to wield in argument, the goal ends up being to find mental stopsigns to be able to dismiss the opponent's argument and totem-like symbols of legitimacy to prop your own arguments up.
I don't know what the way out is, but I can't imagine externalizing our thinking/reasoning even more would be the way.