Meta plans to mark AI-generated content on Facebook, Instagram, and Threads

Meta has recently declared its initiative to tag and identify AI-generated content on its platforms Facebook, Instagram, and Threads, acknowledging the challenges in spotting all AI-created materials. This move follows the spread of controversial AI-generated deepfakes and occurs amidst growing concerns over the handling of artificial images and amended videos, especially with the 2024 US elections on the horizon. Nick Clegg from Meta highlighted the nascent stage of AI-generated content dissemination and expressed Meta’s commitment to adapting its strategies through continuous learning and collaboration with peers and regulatory bodies. Meta aims to set industry standards for identifying such content, working alongside organizations like the Partnership on AI. Despite efforts to use invisible watermarks and other identifiers for AI-generated images, experts like University of Maryland’s Soheil Feizi caution that current watermarking techniques are not foolproof against malicious tampering. Margaret Mitchell of Hugging Face suggests that while watermarking has its limitations, it plays a critical role in promoting transparency by tracing AI content’s origins, thus supporting responsible usage.