Topics
Latest
AI
Amazon
Image Credits:Bryce Durbin / TechCrunch
Apps
Biotech & Health
clime
Cloud Computing
Commerce
Crypto
Enterprise
EVs
Fintech
Fundraising
Gadgets
Gaming
Government & Policy
ironware
Layoffs
Media & Entertainment
Meta
Microsoft
privateness
Robotics
Security
Social
distance
startup
TikTok
Transportation
speculation
More from TechCrunch
event
Startup Battlefield
StrictlyVC
Podcasts
Videos
Partner Content
TechCrunch Brand Studio
Crunchboard
Contact Us
drop a stone and you ’ll likely hit a deepfake . The commoditization of generative AI has led to an absolute explosion of phony message online : According to ID substantiation program Sumsub , there ’s been a 4x increase in deepfakes worldwide from 2023 to 2024 . In 2024 , deepfakes accounted for 7 % of all fraud , per Sumsub , stray from impersonations and account takeovers to sophisticated social engineering science campaigns .
In what it desire will be a meaningful contribution to the fight against deepfakes , Meta is releasing a tool to use imperceptible watermarks to AI - generated video clips . Announced on Thursday , the tool , called Meta Video Seal , is uncommitted in open informant and designed to be integrate into existing package . The pecker joins Meta ’s other watermarking tools , Watermark Anything ( re - release today under a permissive licence ) and Audio Seal .
“ We developed Video Seal to supply a more effective television watermarking solution , particularly for detecting AI - give video and protect originality , ” Pierre Fernandez , AI research scientist at Meta , told TechCrunch in an consultation .
Video Seal is n’t the first engineering science of its kind . DeepMind’sSynthIDcan water line telecasting , and Microsofthas its ownvideo watermarking methodology .
But Fernandez asserts that many existing approaches fall short .
“ While other watermarking tools subsist , they do n’t proffer sufficient robustness to video compressing , which is very prevailing when share content through social platforms ; were n’t efficient enough to break away at scale ; were n’t open or reproducible ; or were derived from image watermarking , which is suboptimal for videos , ” Fernandez said .
In add-on to a watermark , Video Seal can lend a hidden content to picture that can later be uncovered to determine their extraction . Meta take that Video Seal is springy against plebeian edits like blurring and cropping , as well as democratic compression algorithms .
Join us at TechCrunch Sessions: AI
Exhibit at TechCrunch Sessions: AI
Fernandez acknowledge that Video Seal has certain limitations , in the main the trade - off between how perceptible the tool ’s watermarks are and their overall resilience to handling . Heavy densification and important edits may alter the water line or return them unrecoverable , he added .
Of course , the bigger trouble face Video Seal is that devs and diligence wo n’t have much reasonableness to assume it , peculiarly those already using proprietary solutions . In a bid to deal that , Meta is launching a populace leaderboard , Meta Omni Seal Bench , consecrate to compare the performance of various watermarking methods , and coordinate a workshop on watermarking this year at ICLR , a major AI conference .
“ We hope that more and more AI researchers and developers will integrate some form of watermarking into their work , ” Fernandez said . “ We require to collaborate with the industry and the donnish community of interests to progress quicker in the domain . ”
