So how are we going to make sure advanced AIs don't manipulate people on the internet?

I feel like there's a huge threat with AIs pretending to be people online and manipulating and gaslighting people, ruining our social media and forums and communities. What can we do to mitigate that threat?