’s absurd claim that nobody influential would be able to disseminate AI-generated misinformation.
Let’s not let #Facebook once again blind us to the obvious risks to democracy.
These are toy examples. The problem of misinformation due to AI is grossly exaggerated. There's no way to prevent these anyway. LeCun's point is correct. There's too much silly panic. Focus needs to be on regulation of big things and live with the consequences of small things.
Clearly generative AI can be used for misinformation. I do think that there is a large degree of overlap between people who will fall for AI generated misinformation and people who will fall for non-AI generated misinformation, however.
Everyone is in a rush to use these new tools, having zero understanding of their modes of failure.
In this case, having moved so fast, they placed "real" label over fake images. Slander? Violation of election laws?
Move fast and deal with the breakage (aka modern AI safety).