Meta has introduced Video Seal, a new tool designed to combat the growing issue of deepfakes—AI-generated videos—by embedding imperceptible watermarks into them. This tool, which is open-source, aims to help identify AI-generated content, providing a robust solution to the challenge of verifying the authenticity of videos.
Video Seal is part of Meta’s broader effort to combat deepfakes and ensure the integrity of digital content. It integrates into existing software systems and works by embedding a watermark, as well as a hidden message, into videos. The message can later be uncovered to confirm the video’s origin. Unlike some existing watermarking solutions, Meta claims that Video Seal is more effective at withstanding video compression, blurring, cropping, and other common edits, making it more robust for sharing on social platforms.
Also Read: US Probes Meta For Involvement In Drug Trade
While Meta’s watermarking tools are not unique—other companies like DeepMind and Microsoft also offer similar solutions—Meta’s Video Seal is designed to address gaps that exist in other tools, such as poor resilience to video compression and inefficient scalability. However, there are still limitations, particularly when videos are heavily compressed or edited, which may alter or erase the watermark.
Meta is also launching the Meta Omni Seal Bench, a public leaderboard to compare watermarking methods, and plans to organize a workshop on watermarking at the ICLR AI conference later this year to encourage wider adoption of watermarking in AI-generated content. Through these efforts, Meta hopes to collaborate with both the academic and industry communities to advance the development of tools for verifying video authenticity.
