Since generative AI gained traction and astonished pretty much all industries, companies began to realize the money that could be made.
The thing is, AI is advancing fast, and that the results it make starts to become less and less distinguishable if compared to the real thing.
Think about the misinformation that can be generated, the fuss it would create, and the endless controversies that could come after that.
Tech companies know this well, and this is a group of them that are actively developing AI solutions have formed an initiative to address the growing challenge of identifying AI-generated synthetic media and "deepfakes."
Adobe, Microsoft, Intel, and other tech firms, has formed what's called the Coalition for Content Provenance and Authenticity (C2PA), which has a goal of creating an open standard that can certify the origins and provenance of online content.
The C2PA is formed to develop metadata tools that can attach critical information to digital content, including whether AI was used to create or alter it.
This could help content platforms and users identify synthetic or falsified media.
The need for such tools is growing as AI techniques become more advanced, allowing for hyper-realistic fake images and videos.
External pressure is also mounting on tech firms over this issue.
To begin addressing this, the C2PA has introduced a logo that can be attached to certified content.
The logo in question can be attached to certified content, taking the appearance of a symbol on the corner of images and videos, for example.
As the 'icon of transparency,' it's meant to certify real images and brands from forgeries and AI creations.
Created by the super group of companies, the icon is made in response to a world where the images and videos began to feel less trustworthy.
Its two letters, "CR," are a play on "credentials."
The symbol should only appear on content that uses Content Credentials, which is a series of open-source metadata that can be attached to media.
The first to display this symbol, is Adobe.
On content created by Firefly, or through its generative features like Generative Fill, the CR symbol will appear alongside the images.
While the icon doesn’t actually verify if an image is real or AI generated, rather, it's there to only flag if the work has been altered after its creation.
Content Credentials are what the group likens to a “nutrition label” for digital content, explaining the origins of and edits made to an image that you see.
"We think and hope it’ll become as ubiquitous as the copyright symbol, or the Creative Commons symbol,” says Andy Parsons, senior director of the content authenticity initiative (CAI) at Adobe.
The logo itself is also formalizing a standard for a related group, the Content Authenticity Initiative, which is a collection of nearly 2,000 companies, nonprofits, and individuals that have teamed up to promote a shared and open industry standard to authenticate digital media.
While the icon is meant to serve a noble purpose, it's existence comes with a slew of challenges.
First and foremost, the metadata the symbol is based on, is opt-in.
What this means, it's up to the creators themselves to decide whether they want the symbol to be included or not.
The technology that powers the logo have two types of markings: the visible watermark that is added to the intended media, and an invisible metadata.
While the watermark can be erased, like using Content Aware Fill from Adobe, as explained Parsons, the metadata cannot be erased.
This happens because the icon is generated by a site or application.
Because the metadata cannot be erased, it essentially provides an ID to the media. When other apps and websites see an media with this the Content Credentials, they will make this icon appear on or alongside the image through JavaScript.
The logo, and the technology that powers it, is the result of three years of work.
The core intent behind the icon is to battle misinformation and confusion around the origins of media.
"[We’re] hoping it doesn’t focus on authenticity, but transparency," added Parsons. "This is not a blue Twitter check mark that says, 'this is a photo of [whomever].' We want to say ‘this is an authentic photo from a Leica camera.'"
Parsons admit that authenticity and transparency can be confusing, but still, introducing a way to differentiate altered media is needed.
The symbol on media acts like a stamp-of-approval.
It's a way to verify provenance.
While the steps fall short of official regulation, the coalition hopes its standards will promote transparency and trust online.
Following the announcement, the C2PA initiative has released open-source tools that any organization can adopt. Members include media outlets, academics, nonprofits, and tech firms.
