[ad_1]
Throw a stone and you’re likely to hit a deepfake. The commoditization of generative AI has led to an absolute explosion of fake content online: according to identity verification platform Sumsub, there was a 4-fold increase in deepfakes worldwide from 2023 to 2024. In 2024, deepfakes accounted for 7% Of the total scams, according to Sumsub ranging from impersonation and account takeover to sophisticated social engineering campaigns.
In what it hopes will be a meaningful contribution to the fight against deepfakes, Meta is launching a tool to apply imperceptible watermarks to videos generated by artificial intelligence. The tool, called Meta Video Seal, was announced on Thursday, and is available in open source and designed to be integrated into existing software. The tool joins Meta’s other watermarking tools, Watermark Anything (re-released today under a permissive license) and Audio Seal.
“We developed Video Seal to provide a more effective solution for video watermarking, especially for detecting AI-generated videos and protecting originality,” Pierre Fernandez, an AI research scientist at Meta, told TechCrunch in an interview.
Video Seal is not the first technology of its kind. DeepMind’s SynthID can watermark videos, and Microsoft has its own methodologies for watermarking video.
But Fernandez stresses that many current methods are insufficient.
“Although other watermarking tools existed, they did not provide enough power for video compression, which is very common when sharing content across social media platforms; they were not efficient enough to work at scale; they were not open or repeatable; Fernandez: “Or they were extracted from image watermarks, which is not ideal for videos.”
In addition to the watermark, Video Seal can add a hidden message to videos that can later be revealed to identify their origins. Meta claims that Video Seal is resilient against common modifications such as blurring and cropping, as well as common compression algorithms.
Fernandez admits that Video Seal has certain limitations, particularly the trade-off between the tool’s watermark awareness and its overall flexibility for manipulation. He added that extreme pressure and significant modifications may alter the watermarks or make them unrecoverable.
Of course, the biggest problem facing Video Seal is that developers and the industry will have little reason to adopt it, especially those who already use proprietary solutions. In an effort to address this, Meta is launching a public leaderboard, Meta Omni Seal Bench, dedicated to comparing the performance of different watermarking methods, and is organizing a workshop on watermarking this year at ICLR, a major AI conference.
“We hope that more and more AI researchers and developers will incorporate some form of watermarking into their work,” Fernandez said. “We want to collaborate with industry and the academic community to advance faster in this area.”
[ad_2]