Back

Fakes Aren't Funny -- or Are They?

The tools to create fake images are in the hands of the masses! From PhotoShop to InDesign to DeepFace Lab and Zao, you can make anyone say anything. Want a free drink on the plane? A simple editing on your self-printed boarding pass might do the trick (don't try this please!).

This talk will explore the ways in which images/video/documents can be manipulated, including some humorous examples. It will also discuss the serious aspects (e.g. lots of insurance fraud now involves faked "photo evidence" of flooded basements, etc.) As we move to elections in the U.S. and Canada, it is inevitable that we'll see political deepfakes and other chicanery.

After establishing that fakery is child's play, the talk will discuss various different ways of detecting fakes. Everything from checking the carotid pulse of a speaker to background matching and phoneme analysis. The author has developed a blockchain-based validation model that would, for example, have proven which "CNN Jim Acosta video" was the original and which was the fake that the White House tweeted out.

Finally, this talk will give an important message to all of us who have hacker abilities: just because you can do it doesn't mean you should!