This AI Can Detect Image Manipulation, And Might Well Be Our Savior

Much has been written and discussed about the threat of image manipulation on our society. As deepfakes and AI-powered image editing continue to grow in usage, it’s only a matter of time before seeing is no longer believing.

The ramifications of this would be catastrophic. Edited videos of politicians could alter the outcome of elections, celebrities could be manipulated into awkward situations that never truly happened. Lives could be ruined.

It’s probably safe to say the biggest driver behind the spread of manipulated images will be social media platforms. Unfortunately, they don’t appear to be taking the threat too seriously.

Bill Posters and Daniel Howe, with the help of ad agency, Canny, recently posted a deepfake of Mark Zuckerberg himself on Instagram. Despite the fact that the creators announced that it was fake, Instagram said they would only remove it once a third-party had verified it as manipulated.

The slowness in responding to deepfakes on their platform suggest Instagram (and other platforms) don’t quite understand the issue.

read original article at——artificial_intelligence-5