As social media platforms move to crack down on deepfakes and misinformation in the US elections, an Indian politician has used artificial intelligence techniques to make it look like he said things he didn’t say, Vice reports. In one version of a campaign video, Manoj Tiwari speaks in English; in the fabricated version, he “speaks” in the Hindi dialect of Haryanvi.
To be honest my opinion is that as long as it translates into the same thing he is saying in the original language, it is not "things he did not say". It is an accurate translation just like captions are. Of course if he were to say he speaks Haryanvi that would be a lie... and no politician would want to be caught out lying.
But it does show there can be positive uses for AI, and of course that it is going to get very difficult to really know what is original or not. Knowing what we know about behaviour on the Internet (and within politics) I can only imagine where this is going... but the barn door is now opened and unfortunately there is no going back.
See An Indian politician used AI to translate his speech into other languages to reach more voters
How intentionally faked videos might affect US elections is unclear