Protect images from fake video creation by disrupting AI reading process
LETHE: Defending against Unauthorized Image-to-Video Generation via Value Diversity Collapse
Computer Vision and Pattern RecognitionCryptography and Security
Summary
Turning a single photo into a fake video can be misused to make things look real when they are not. The authors study how AI models use parts of an image to create videos and find that making the image’s details less varied can confuse these models. They create a new way to subtly change images so any video made from them looks wrong or corrupted. This method, called LETHE, works well on modern systems and can help protect personal photos from being turned into unauthorized videos.
What this means in practice
- •For photo sharing platforms: Add image protections that prevent users' photos from being used to generate fake videos without consent.
- •For social media companies: Implement defenses to reduce misuse of personal images by unauthorized AI-generated videos in user content.
Authors
Boheng Li, Yueming Xing, Yiming Li, Jie Zhang, Tianwei Zhang
Abstract
Image-to-video (I2V) diffusion models can animate a single reference image into high-quality, subject-consistent videos, posing threats of misuse as anyone can turn personal photos into fabricated videos without consent. A promising solution is to proactively add an adversarial perturbation to the image before release, such that videos generated from it would be unfaithful or visually corrupted. However, due to an insufficient understanding of what governs how subsequent frames read the reference image in modern I2V models, existing protections struggle to offer strong protection under common perturbation budgets. In this paper, we approach the problem by exploring this underlying reading mechanism. Through a convex geometric analysis, we theoretically identify the spread of the reference value vectors as a key factor governing this reading process, and empirically find that even a mild contraction of this spread can severely degrade the generated video, making it an efficient lever for protection. Guided by this insight, we propose LETHE, a novel protection method that directly disrupts this reading process. LETHE pairs two complementary objectives, where a value collapse objective drives the reference value vectors toward uniformity, while an attention attraction objective draws subsequent frames toward these collapsed tokens to amplify corruption. We further construct two human-centric I2V protection benchmarks and conduct extensive experiments with multiple modern I2V models, which demonstrate the state-of-the-art protection performance and versatility of LETHE.