Some Instagram users have recently noticed a ‘Made with AI’ tag appearing in the top left of reels and posts. In many cases, the tag is appearing on real photos and not images made using an AI image generator as the tag implies, and photographers aren’t happy about it.
So what is going on? Instagram users @colinandsamir posed the question to the Head of Instagram Adam Mosseri (@mosseri) to clear things up (see below) and he said:
“It’s just the metadata on the image. People are doing benign things in Photoshop that are technically using AI. And then we’re just respecting the file’s mark saying ‘Hey, AI was used to make this.’ We have to figure this out because clearly this is frustrating and not the ideal experience.”
Disgruntled photographers have not taken kindly to the suggestion that their images are fake, but are instead real photos that are edited using Adobe Photoshop which has improved AI editing tricks.
For some creators such as Peter McKinnon, Instagram’s binary tagging approach feels like an overreach.
Matters get complicated and fast: Instagram is also adding the ‘Made with AI’ tag on many AI-generated images and ultimately expecting viewers to use their own judgement as to what’s real, so we’re back at square one.
In a time when AI will make it almost impossible to know the truth, Instagram’s degree of caution is understandable but that’s of little comfort to photographers sharing real albeit edited images that are being tarred with the same AI-generated brush.
And, as Mosseri goes on to say, there continues to be content online that is generated by AI that is not tagged to that effect and could therefore be assumed as being real, “There will be content on all platforms that is made with AI and is not marked with AI because the person who made it specifically tried to avoid that.”
Handling AI content online remains a complex task.
What else can Instagram do?
Instagram scans the metadata of an image to ascertain if it’s real or AI-generated. Those savvy enough to upload an image without metadata – for example, a screenshot – can circumvent Instagram’s tag with their fake images.
Meanwhile, a photographer who used Photoshop’s content-aware tools to remove a blemish on skin of a real person or a stain on clothing could be lumped with the ‘Made with AI’ tag if those changes are in the image metadata. Clearly, the situation is far from ideal, but it’s also far from straightforward to navigate.
Instagram users have been responding to Colin and Samir’s Instagram posts with mixed feelings and their own ideas of how Instagram should point out fake images.
User @voust says, “There should be ‘edited with AI and made with AI’… ‘made’ kind of implies the whole image is artificial”.
@voust suggests an alternative to Instagram’s binary approach for photographers for whom authenticity matters: it could scan real photos with Adobe‘s Content Credentials metadata that proves authenticity, along with if there were any edits made to the image to avoid the ‘Made with AI’ tag, and that an ‘edited with AI’ tag is acceptable.
Photographers have been faking and manipulating photographs ever since the invention of photography, but we’re on new AI ground that adds a whole other layer. Photoshop’s Generative Fill AI tools are pushing what’s possible in its flagship editor, and confusing what’s real.
In photography circles, AI-generated images aren’t seen as photographs, nor is replacing an entire sky in your photo using Photoshop’s Generative Fill, while light editing of real photos is on the whole acceptable – and it’s verging on insulting that these edited photos are being tagged as ‘Made with AI’.
A single ‘Made with AI’ tag leaves too many questions and in some cases unfair suspicion. To a degree, ‘edited with AI’ also leaves questions: how much editing has taken place; an entirely new landscape around your subject or simply a small piece of litter removed from the ground?
I think Instagram is on the right track, but I also hope that it listens to photographers and finds a more acceptable way forward.
