It is undeniable that today the Artificial Intelligence (AI) It has infiltrated almost every aspect of our lives. Since the boom experienced with ChatGPT by OpenAIthe public's fascination (and obsession) with this technology has skyrocketed, leading large companies like Manzana, Goal and Google integrate AI into their platforms. What has come to detonate this strange panorama where this technology at times feels omnipresent.
In fact, among the most popular tools powered by Artificial Intelligence we find those like Midjourney, DALL-E and similar that can generate realistic images and videos, which has further blurred the line between real and virtual.
Where on more than one occasion we have come across digital pieces that it is impossible to distinguish if they captured something real or if they were assembled by AI. Not long ago, to combat this situation, the people of Meta began tagging in Facebook, Instagram and Threads, those images that in theory would have been created with these tools.
But it seems that never worked.
Meta fails to detect which images were created by Artificial Intelligence
The purpose of these labels was to help users discern between AI-generated content and real photographs. However, in recent days the system's inability to accurately label images has been exposed.
According to an extensive report by colleagues from TechCrunchMeta has been applying the “Made with AI” label to images that were not created with said technology and in fact they themselves verify it, capturing cases where even photos captured analogically decades ago and digitized to be shared on social networks are identified erroneously as created by AI.
![Meta labels real photos as generated by Artificial Intelligence on Facebook and it is Adobe's fault Meta implemented a function on Facebook to “detect” when an image was created using Artificial Intelligence, but it does not work.](https://www.fayerwayer.com/resizer/v2/PAAPZ5T6IRBGFB52EKLNFQVAJQ.png?auth=7dcc7b4acec6cc52cec515ebd551e784361901f9b81a6a88e6d38a8028f8aa86&width=800&height=1234)
Even the person writing these lines can attest that he has experienced a similar case, where a retouched photograph in Adobe Lightroom, Without any AI intervention, it has still been mislabeled.
This recurring phenomenon has raised concerns among photographers, especially when their original images are inappropriately tagged. Although there seems to be an obvious explanation for the origin of the problem.
Adobe could be to blame for Meta misidentifying AI-generated photos on Facebook
The former White House photographer, Pete Souza, attributes this problem to changes in Adobe's editing programs, particularly in the cropping function, which apparently would implement some kind of AI technology to improve outlining, but that would automatically mark the image as not created by a human being.
Image editing has become increasingly complex, especially with the arrival of Artificial Intelligence tools that allow you to modify images with a single click. While some of these edits may be subtle, they are still manipulations made with the controversial technology.
![Meta labels real photos as generated by Artificial Intelligence on Facebook and it is Adobe's fault Photo mislabeled by Meta. Not AI-created.](https://www.fayerwayer.com/resizer/v2/CQWPGEJ2UNAHHBTVLV7DNSU5RQ.png?auth=fe8cc5f9d8a4d7506974746eb212ad2a22a2272b83a0174ce523d2bf38064ad5&width=800&height=797)
Goalaware of this situation already has declared through his official blog that it is actively working with companies to improve their labeling process and that its “intention has always been to help people know when they see content created with AI.”
However, for the time being, non-working tags will remain, meaning images edited with AI tools like those in Adobe's suite could be tagged incorrectly.