In a world where AI-generated art is becoming increasingly prevalent, a recent study has revealed a troubling truth: existing tools designed to protect artists’ work from unauthorized use are not as effective as hoped. Researchers have shown that popular protection tools like Glaze and NightShade can be easily bypassed, leaving creators vulnerable to having their copyrighted material used without consent in AI training datasets.

The study, led by a team of international researchers, introduced LightShed, a new method that can detect and reverse-engineer the distortions applied by these protective tools. With an astonishing accuracy rate of 99.98%, LightShed effectively strips away the protections, rendering the images usable again for generative AI models. This revelation underscores the urgent need for more robust and adaptive defenses in the rapidly evolving landscape of AI and digital creativity.