Meta, the parent company of Facebook, plans to label images generated by artificial intelligence (AI) products, including those from OpenAI and Midjourney, according to Nick Clegg, Meta’s President of Global Affairs.
The labels will be implemented on Facebook, Instagram, and Threads in the coming months, aiming to differentiate between AI-generated and human-created content.
Even Meta’s own AI-generated images will carry labels. The move comes amid concerns about the blurring line between human and synthetic content. However, Clegg acknowledges the labels are not a perfect solution given the scale and complexity of AI-generated content.
Meta also plans to introduce a feature allowing users to voluntarily label AI-generated audio or video when uploading to address the challenge of identifying such content.
Recent incidents, including the viral spread of AI-generated explicit images of Taylor Swift and a fake robocall impersonating President Biden, highlight the risks associated with AI-generated content.
With elections approaching, Clegg emphasizes the industry’s responsibility to provide visibility and help users distinguish between synthetic and authentic content. Meta intends to continue labeling AI-generated images through the next year, particularly during crucial global elections, to assess the effectiveness of their efforts.
Clegg supports legislation regulating AI but does not explicitly comment on a Senate bill proposing a ban on deceptive AI content in political ads.