Adobe Wants to Make Photoshop a Tool For Spotting Fake Photos – Gizmodo

G/O Media might get a commission.

The usage of cryptography does make it hard for the CAI metadata embedded in the images to be tampered with, but its not impossible. The CAI system, at least in its current form, doesnt consist of any safeguards to avoid people from taking screenshots and then modifying the signed and authorized images. Support for the CAI system has to be baked into digital video cameras, computers, mobile devices, and any platform that can be utilized to share images en masse.

Using cryptography does make it hard for the CAI metadata embedded in the images to be tampered with, but its possible. Theres also the capacity for the metadata to be totally stripped off an image and changed with phony information. The CAI system, at least in its existing type, does not include any safeguards to avoid individuals from taking screenshots and then modifying the signed and licensed images. One day such safeguards might be built into an os, limiting the ability to screenshot an image based upon its CAI qualifications, but thats a long way off.
On the planet of digital photography, Adobe carries a lot of weight and impact, and it will need to leverage that as much as it can for its Content Authenticity Initiative to have any hope of being an effective tool versus fakes. Getting a handful of huge name newspapers on board simply isnt enough. Support for the CAI system has to be baked into digital cams, computers, mobile gadgets, and any platform that can be used to share images en masse. And it requires to be combined with a huge push to not just educate users on how to use this info to identify phony news, but also a desire to in fact take a couple of extra seconds to verify on their own if an image is genuine or not– and that may be the biggest difficulty. If the pandemic has actually taught us anything, its that individuals are excited to believe anything that supports their own suitables, no matter what the experts may state.

For 30 years, smart pixel-pushers have been using Photoshop to edit and manipulate images, and now that computer systems can develop doctored pictures all on their own using sophisticated AI, Adobe wishes to leverage its image-editing tools to assist validate the authenticity of photos.

The CAI system does have the prospective to help slow the spread of disinformation and manipulated photos, but it will require users to have access to the protected metadata, and to take the effort to verify that an image theyre looking at is authentic. For instance, pictures of a supposed violent protest shared the next day on Facebook might be quickly debunked by the metadata exposing the images were actually shot years earlier in another part of the world.
For it to be reliable, the CAI method has to be extensively accepted and executed on a big scale by those creating photo content, those releasing it, and those consuming it. Photographers working for official news organizations could easily be needed to use it, however thats a small sliver of the images being uploaded to the internet daily. For the time being, it does not seem like social networks giants such as Twitter or Facebook (which owns Instagram) are preparing to jump on the CAI bandwagon, which is problematic provided thats where a lot of so-called “phony news” gets published and extensively shared now.

As Wired reports, Adobe partnered with a handful of companies in 2015 to assist establish its Content Authenticity Initiative: an open standard that stamps photographs with cryptographically protected metadata such as the name of the photographer, when the image was snapped, the precise location an image was taken, and the name of editors who might have controlled the image in some manner along the way, even if its just a fundamental color correction.
Later on this year, Adobe plans to include the CAI tagging capabilities into a sneak peek release of Photoshop so that as images are opened, processed, and saved through the app, a record of who has actually dealt with or controlled the image can be continually documented in an ever-growing log developed right into the image itself. The CAI system will also include information about when a photo was published to a news website like The New York Times (one of Adobes initial partners on this initiative in 2015) or a social media network, such as Adobes Behance, where artists and users can easily share their productions online.

Screenshot: Gizmodo (Other).