AI-Powered Facial Mask Removal Is Not Suitable For Identification
Emily A Cooper, Hany Farid

TL;DR
This paper evaluates the reliability and risks of using AI to unmask faces in images, highlighting its unsuitability for accurate identification due to potential misidentification issues.
Contribution
The study provides a large-scale analysis of commercial AI facial unmasking tools, assessing their effectiveness and risks in real-world identification scenarios.
Findings
AI unmasking often fails to reliably match faces to true identities.
Generated unmasked faces can lead to misidentification and false positives.
The technology poses significant risks for misuse in criminal investigations.
Abstract
Recently, crowd-sourced online criminal investigations have used generative-AI to enhance low-quality visual evidence. In one high-profile case, social-media users circulated an "AI-unmasked" image of a federal agent involved in a fatal shooting, fueling a wide-spread misidentification. In response to this and similar incidents, we conducted a large-scale analysis evaluating the efficacy and risks of commercial AI-powered facial unmasking, specifically assessing whether the resulting faces can be reliably matched to true identities.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
