Image: Cath Virginia / The Verge, Chris Strider As the US presidential election approaches, the web has been filled with photos of Donald Trump and Kamala Harris: spectacularly well-timed photos of an attempted assassination; utterly mundane photos of rally crowds ; and shockingly out-of-character photos of the candidates burning flags and holding guns. Some of these things didn’t actually happen, of course. But generative AI imaging tools are now so adept and accessible that we can’t really trust our eyes anymore.
Some of the biggest names in digital media have been working to sort out this mess, and their solution so far is: more data — specifically, metadata that attaches to a photo and tells you what’s real, what’s fake, and how that fakery happened. One of the best-known systems for this, C2PA authentication, already has the backing of companies like Microsoft, Adobe, Arm, OpenAI, Intel, Truepic, and Google. The technical standard provides key information about where images originate from, letting viewers identify whether they’ve been manipulated.
“Provenance technologies like Content Credentials — which act like a nutrition label for digital content — offer a promising solution by enabling official event photos and other content to carry verifiable metadata like date and time, or if needed, signal whether or not AI was used,” Andy Parsons, a steering committee member of C2PA and senior director for CAI at Adobe, told The Verge . “This level of transparency can help dispel doubt, particularly during breaking news and election cycles.”
But if all the information needed to authenticate images can already be embedded in the files, where is it? And why aren’t we seeing some kind of “verified” mark when the photos are published online?
The problem is interoperability. There are still huge gaps in how this system is being implemented, and it’s taking years to get all the necessary players on board to make it work. And if we can’t get everyone on board, then the initiative might be doomed to fail.
The Coalition for Content Provenance and Authenticity ( C2PA ) is one of the largest groups trying to address this chaos, alongside the Content […]
This system can sort real pictures from AI fakes — why aren’t platforms using it?
