Meta, the parent company of Facebook and Instagram, is planning to identify and label images produced by third-party artificial intelligence services. The initiative, revealed by Nick Clegg, Meta’s President of Global Affairs, involves embedding invisible markers into files, alerting users to distinguish digital creations from authentic photos.
While Meta already labels content generated through its AI tools, this expansion aims to encompass images from external providers, including OpenAI, Microsoft, Adobe, Midjourney, Shutterstock, and Alphabet’s Google.
This strategic move aligns with industry efforts to establish standards for handling the challenges posed by generative AI technologies, capable of generating convincing but fake content. In an interview with Reuters, Clegg expressed confidence in the reliability of labeling AI-generated images, acknowledging that audio and video content pose greater complexities.
Despite ongoing development in these areas, Meta plans to implement penalties for individuals failing to label manipulated audio and video content. Interestingly, Clegg mentioned the absence of a viable mechanism to label AI-generated written text, stating, “That ship has sailed.”
The announcement also leaves uncertainty about whether Meta will apply similar labels to generative AI content on its encrypted messaging service, WhatsApp.
This development follows Meta’s independent oversight board’s recent criticism of the company’s policy on misleadingly doctored videos, prompting Clegg to agree that the existing policy needs adjustments.
He views the labeling partnership as a proactive step toward addressing the challenges posed by the surge in synthetic and hybrid content.
