Deepfakes Are Evolving. This Company Wants to Catch Them All
Deepfakes Are Evolving. This Company Wants to Catch Them All
Hany Farid, a leading expert on image and video manipulation, says that detecting deepfakes will take more than AI alone.
Non-clickbait title:
UC Berkeley Professor Founds a Deepfake Forensics Company GetReal Labs.
So AI is Pokémon now
I've said before, what's needed here is an independent org or maybe even multiple that can provide certification that pieces of media were created without the use of generative tech, and then open up companies that publish anything missing the cert to lawsuits from the public.
Verify then trust publishing would go a long way to contain misinformation in general, not just deep faking and other generated content.
Theoretically good but it'd also be yet another way to raise the bar to entry for all creative industries giving buffalo buffalo buffalo established players an even bigger advantage.
Nah just get the work certed, it'd be available for anyone, it just needs to pass the "No AI was used at all" test, and that can be done by giving creative software tools the ability to sign files whenever AI or screen shotting tools get used in their making
I think the approach of providing software tools the ability to grant the cert if no AI tools are used will be the way to really spread the practice, can even work with them to create a cert mode that just doesn't even present those tools to the user in the first place.
The next step would be getting publishing sites to use the cert or signature as a check that either blocks the upload or more leniently get a visual marker that indicates it was not made with generator tools.
Plus checking for the cert also eliminates the crisis of teachers having to outsmart GPT to avoid their students cheating on takehome essays.