Deepfakes vs fake detection. The arms race is on.

Few posts down the line I was writing on how the rise of the deepfake technology will require strong countermeasures. The deepfakes are becoming so precise that it is almost impossible to distinguish them from the real video by naked eye.

If you do not realise it is already happening, just have a look at the recent TikTok phenomenon, DeepTom Cruise; https://www.youtube.com/watch?v=wq-kmFCrF5Q&ab_channel=VFXChrisUme

So now, we will need support of an AI in order to be able to distinguish fakes from a real videos.

That, in turn, will lead to more sophisticated & undetectable deepfakes. And so the spiral begins.

“While in the beginning, AI-generated synthetic videos (using DNNs — Deep Neural Networks) were designed to deceive the human eye from authentic videos, advancements in technology are now taking measures to produce deepfakes undetectable to detection algorithms. “

Source: https://futureofmarketinginstitute.com/deepfakes-will-influence-the-future-of-marketing-whether-we-want-it-or-not/

The good guys will fight back with tools like the one eye reflection analysis software that is almost bulletproof.
Almost… because there is still that 6% window where fakes can remain undetected.

New Deepfake Spotting Tool Proves 94% Effective – Here’s the Secret of Its Success

My prediction?

You will soon be enabling anti-fake plugins while surfing one youtube, TikTok or other platforms, same way you switch on your ad-blockers. Possibly some platforms will have those integrated in them & analysis will occur upon uploading the user generated content.

In any case – deepfake vs deepfake detection algorithms struggle will become part of our daily online experience & inseparable part of common cyber security measures.