Meta has been labeling pictures created using artificial Intelligence on Facebook, Instagram, and Threads to distinguish real content from AI-generated media. The company has been actively pursuing watermarking since May and has lately drawn criticism from photographers for incorrectly tagging the original or real pictures as AI. A recent report states the reason for the mistake could be exceptionally advanced editing tools that make it impossible to tell the difference.
Many photographers are sharing the incorrect AI tagging by Meta on pictures they have taken themselves
According to a report by TechCrunch, Meta's pursuit of identifying real from AI-generated pictures has landed it in hot waters. Several users, particularly photographers, have pointed out the mislabeling of their original content by the company. The photographers shared their stories of photo-marking and one White House photographer Peter Souza said that a picture he took of a basketball game was AI watermarked.
Another incorrect label that made quite the buzz was the picture of the Indian Premier League Cricket Tournament being AI-marked. It is said the tags appear when the content is viewed on the phone and does not show up on the web. When users try to edit or alter the label, there is no option to do so, creating a hurdle for original content creators in their creative process.
The White House photographer did mention that he edited the pictures through Adobe's photo editing tool before posting them, which might have resulted in Meta detecting them incorrectly. Many content makers expressed their dismay over minor tweaks in images, leading to them being AI watermarked.
One photographer took his frustration out on Threads, saying how the company's labeling method has no value if retouched pictures are going to be unfairly tagged. TechCrunch also emphasized how tools such as Generative AI Fill that can remove unwanted objects from pictures can make it extremely hard for AI detectors to tell them apart correctly.
Meta spokesperson Kate McLaughlin told The Verge that the company was aware of the ongoing issue and is working on refining the labeling more by specifying the amount of AI used in creating the media. She further said:
We’re actively working with these companies to improve the process so our labeling approach matches our intent.
While the AI frenzy is not stopping any time soon, it is becoming increasingly cumbersome for companies to establish the real from the slop. The metadata marking to show the amount of alteration in the images could help clear the air of confusion.
Follow Wccftech on Google to get more of our news coverage in your feeds.
