Artists Gain Upper Hand Against AI with Innovative Data Poisoning Tool
Originally Published 2 years ago — by Engadget
Nightshade, a new tool developed by University of Chicago professor Ben Zhao and his team, allows artists to add undetectable pixels to their work, corrupting AI's training data and protecting their creations. This comes as major companies like OpenAI and Meta face lawsuits for copyright infringement. Nightshade alters how machine-learning models produce content, potentially causing them to misinterpret prompts or generate different images. The tool follows the release of Glaze, which also alters pixels but makes AI systems detect the initial image as something entirely different. While Nightshade could encourage proper compensation for artists, it also raises concerns about potential misuse.
