This is too easy, I struggled a bit with no sound but video to detect real or fake because the lighting can be a bit weird but for the voices you can hear it in a split second that they're fake
Agreed. The video and text fakes are getting better, but the audio (at least here) isn't close. It's not even an uncanny valley, it's worse than an SNL parody.
In cases without audio, watching the lips for sync was also a reliable giveaway.
I only did about half of the samples, but things became harder further in. I'm not sure if they were better done or I was just becoming bored and losing focus though.
