Meta To Label AI-Generated Content On Facebook And Instagram

Meta has announced that they will be addressing the rising deepfake issue on their social media platforms, Instagram (including Threads) and Facebook. They mentioned in a release that they will be doing so by labeling AI-generated content as such, under posts.

Nick Clegg, the President of Global Affairs at the company commented, “People want to know where the boundary lies between human and synthetic content.” As a response to user feedback, Meta has already been labeling photorealistic images generated using its Meta AI feature as “Imagined with AI.” Now, the company aims to extend this labeling practice to AI-generated content from other companies as well.


A Deeper Dive Into Meta’s Decision


The main reasons for Meta’s decision to label AI-generated content, as stated in Meta’s release, are:


  • The increasing prevalence of AI-generated images online, some of which are harmful, such as fake explicit images or politically motivated misinformation.
  • Legal obligations, as laws are being passed that require online platforms to take more responsibility for the content shared on their sites. For example, the UK’s Online Safety Act makes it a crime to upload fake explicit images without consent.


  • The need for transparency and clarity about which images are AI-generated and which are not, to help users understand the content they interact with online.


  • The anticipation of similar actions from other companies in response to their initiative, setting a standard in the industry.


  • The hope that this action will lead to a decrease in harmful fake images online, although the effectiveness of this measure remains to be seen.


  • The potential legal and financial consequences if they fail to adequately address this issue.



Collaboration for Standardisation


“We’re working with industry partners on common technical standards for identifying AI content,” stated Clegg. By aligning with technical standards developed through forums like the Partnership on AI (PAI), Meta aims to ensure consistency in identifying AI-generated images across various platforms.


Implementation of Detection and Labeling Mechanisms


Meta is making use of advanced tech to detect and label AI-generated images effectively. By embedding invisible markers within image files, such as IPTC metadata and invisible watermarks, Meta aims to enhance the accuracy of its detection mechanisms.

The company also plans to introduce features that allow users to disclose when they are sharing AI-generated content, the way TikTok does, for the application of appropriate labels. This empowers users as well, to contribute to the transparency of the content they share while facilitating Meta’s efforts to accurately identify and label AI-generated images.


The Challenges And Developments In The Future


Meta acknowledges the ongoing evolution of technology and the need for continuous adaptation. Sir Nick Clegg said, “It’s not yet possible to identify all AI-generated content,” given the digital world’s dynamic nature.

To stay ahead of such challenges, Meta is actively exploring innovative solutions, such as developing classifiers to detect AI-generated content even in the absence of invisible markers. Additionally, research initiatives like the development of Stable Signature watermarking technology aim to bolster the resilience of detection mechanisms against adversarial threats.

As Sir Nick Clegg stated, “Generative AI tools offer huge opportunities, and we believe that it is both possible and necessary for these technologies to be developed in a transparent and accountable way.”