A new collaborative study by researchers from UTSA, University of Cambridge, and TU Darmstadt revealed that current digital art protection tools — Glaze and NightShade — can be bypassed by a new method called "LightShed". On June 23, 2025, the institutions confirmed via press releases that LightShed detects and removes poisoning perturbations with 99.98 % accuracy, restoring images to a state suitable for AI model training. UTSA Today and TU Darmstadt sources corroborate the results. This finding underscores the insufficiency of technical defenses alone and reinforces calls for more robust adaptive protections and legal frameworks.
New "LightShed" Method Neutralizes AI Art Protection Tools Like Glaze and NightShade
