Meta, the company behind Facebook and Instagram, is overhauling its policies on manipulated media in an effort to combat deceptive content, particularly content created using artificial intelligence (AI). This comes just months before the crucial US elections in November, where experts warn AI-generated content could pose a major challenge.
The new policy, announced by Meta’s Vice President of Content Policy, Monika Bickert, introduces “Made with AI” labels for content starting in May. This expands on Meta’s previous policy which only targeted specific types of manipulated videos.
Additionally, Meta will implement clear and prominent labels for manipulated content deemed a high risk of misleading users on important issues, regardless of whether AI was used. These “high-risk” labels will go into effect immediately.
This shift in strategy moves Meta away from simply removing manipulated content, instead opting to keep it accessible while informing users about its creation. The company previously announced plans to detect AI-generated images using invisible markers, though a start date hasn’t been revealed.
The new labeling policy applies to Facebook, Instagram, and Threads, but not to other Meta platforms like WhatsApp and Quest VR headsets.
These changes come in response to concerns about the potential misuse of AI in the upcoming elections. Political campaigns have already begun utilizing AI tools, raising questions about adherence to guidelines set by Meta and leading AI companies like OpenAI.
The update follows criticism from Meta’s own oversight board in February. The board deemed the existing policies “incoherent” after reviewing a manipulated video of Joe Biden on Facebook. The video, which remained despite misleading viewers, highlighted a gap in Meta’s approach that only addressed AI-generated content or manipulated speech. The board urged Meta to expand its policy to include all manipulated content, including audio and videos portraying actions or statements never made by the individuals depicted.
