how to combat generative AI deepfakes on blockchain—comments from Steinberg
Generative AI deepfakes of the Pentagon fake explosion photo or the Pope in the white puffer jacket is only the beginning. How do we combat it? By verifying content on the blockchain which will help the public know if a photo or video is real or fake.
Lou Steinberg, founder of cyber research lab CTM Insights, says there will be technology solutions within a few months that will detect generative AI deepfakes in real time whether it be from voice, video or pictures.
Steinberg says while most are trying to protect bits and bytes the better solution is to assign signatures—on individual words or images— in real time and publishing on the blockchain to combat synthesized photo and videos that can be created with generative AI.
For instance, a video or Zoom of a CEO can be turned into text in real time which has also been verified by the creator and put on the blockchain. The public and media would know instantly if a business leader or politician actually said those all or some of words. The same technology would apply to photos.
Some are developing technology to create a ’wax’ seal to authenticate entire videos and pictures. Steinberg says if you take a soundbite or snippet out of the the middle and aren’t using entire package then that type of solution can’t be verified like it can on blockchain technology.