The use of AI to produce content on social media is now commonplace and we are starting to see AI content being used for fraudulent purposes. Two weeks ago, Siti Nurhaliza's voice was used in the advertising of products that were supposedly certified by her. Meta has announced they will start labeling AI-generated content on Facebook, Instagram and Threads in the near future.
The label says images generated by AI will be visible in these applications. When an icon is pressed, information about what was used to generate it will be visible to all. Meta will support several image identity authentication systems such as C2PA and IPTC. This system is used by Google, OpenAI, Microsoft, Adobe, Midjourney, and Shutterstock to place meta data on the identity of images and label them as generated by AI.
Currently only AI-generated images contain easily visible metadata and watermarks. For content producers who generate audio and video with AI, the option will be given to state that the content is produced with the help of AI to avoid audience confusion. Failure to place this label may subject them to action.
All of these features were announced by Meta in preparation for the US Presidential election which is expected to see the use of AI increase dramatically. Fake content that used to be just pictures now exists in the form of video and audio.