Honestly this scenario is so terrifying to me that I almost hate to even discuss it. What happens when people can make convincing fakes like that of religious figures and weaponize it to incite violence or install authoritarian governments?
How much more powerful and devastating would calls for genocide be, if driven by deepfake videos? Especially in parts of the world where people wouldn't even be a little bit skeptical of something outrageous that they saw with their own eyes.
Just look at the current political climate in the US. How much more polarized and volatile would things be if someone could fake videos showing politicians abusing children or planning a violent 'government takeover'.
If/when this tech gets good enough, it will absolutely be disruptive to civilization.
You do realize it was already possible to do this? What's new here is not the capability to create a convincing fake video, but the deep learning/AI aspect of it.
>How much more powerful and devastating would calls for genocide be, if driven by deepfake videos? Especially in parts of the world where people wouldn't even be a little bit skeptical of something outrageous that they saw with their own eyes.
>Just look at the current political climate in the US. How much more polarized and volatile would things be if someone could fake videos showing politicians abusing children or planning a violent 'government takeover'.
Probably not much. There aren't a lot of people out there who would commit mass murder or political violence if only there were convincing documentary evidence, but who wouldn't otherwise.
No, it really won't, unless it involves them personally. If a celebrity or politician denies it was them in "that video," most people will believe whatever conforms to their biases about that person.
This deepfake situation, and a similar situation that will occur by synthesizing fake speech in people's own voices, is unstoppable. All the complaining in the world won't stop it, because the barrier to entry will soon be trivial. What is the proposed response? Criminalize running certain kinds of neural nets? Or social media shaming anyone who publishes any fake (that's detectable)?
We all just need to get used to the fact that pretty soon audio and video recordings of people won't be verifiable without careful checking and comparing, and eventually maybe not even then.
Look at the bright side: someone can have a recording of you saying something crude and disrespectful and describing sexual assault, and if you're running for president you won't even have to admit it was you anymore, because soon anyone will be able to fake that kind of thing. Video's more difficult, but that too might happen.