from the ai-and-elections dept

It seems that every election cycle over the last decade has had concerns being raised about the state of “political deepfakes” and their ability to impact the election. So far, these have always been overblown, and there’s been little evidence of deepfakes having any real impact at all — perhaps other than creating the concept of “ the liar’s dividend ” enabling politicians caught doing actually bad things to falsely claim that it’s “fake news” and a “deepfake.” That could change, of course, and there are already concerns being raised about the impact of deepfakes midterms along with the fact that most of the major models are letting people create deepfake content.

The concern raised in the Tech Policy Press piece is less about putting a politician in a compromising photo or video (those don’t seem to get very far before people recognize they’re fake) and more about using these tools to produce ancillary artifacts that help support claims of other conspiracy theories:

Every…