Its a combination of misinformation and hyperpartisanship that's the risk
When there is no news source that both sides will trust, how do you counter the misinformation? Will the side that benefits be willing to give up that benefit by using their news outlets to tell people it is misinformation? How does that even work for people who don't listen to "news" at such at all, but rely only on what their friends share via social media? Someone who shares something false may not want to admit they were duped and share a retraction/correction. Some are so unwilling to admit fault they will double down and attack anyone who dares to correct them.
The other problem is that AI is good enough (or many people BELIEVE is good enough) that we will see video evidence of a candidate making a specific statement denying it and claiming it was an AI fake. Won't matter when people stand up and say "yes he did say it I was there" because the candidate having got out in front of it by making the denial will be believed by his followers regardless of how much evidence to the contrary is piled up.
We've already seen this in fact, when the pussy grabbing tape was released, Trump at first claimed it was a fake. Though Trump does have a pattern where at first he denies something, then later admits and it claims it was no problem (i.e. "perfect call", "the president records act lets me have those documents", etc.) so he might have trouble maintaining a claim of an AI fake for long if he follows his usual pattern.