Does that matter when they stakes are as high as these arguments always claim?

If we're doomsaying about a "post-truth society", we're talking about high-stakes society-scaled skullduggery.

If you're aiming for that level of disruption, easy deepfakes vs hard video/photo editing is not an issue, getting people to trust your made up chain of custody is.

-

This is like when people worry about general AI becoming self-aware and enslaving mankind... the "boring version" of the danger is already happening: ML models being trained on biased data are getting embedded in products that are core to our society (policing, credit ratings, etc.), and that's really dangerous.

Likewise, people worry about being able to easily make fake news, when the real danger is people not being equipped to evaluate the trustworthiness of a source... and that's already happening.

You don't even need a deepfake, you tweet that so and so said X, write a fake article saying they said X, amplify it all with some bots, and suddenly millions of people believe you.