Court Exhibit Names Artists Midjourney Scraped To Train Its AI; Includes Many Hugo Winners - File 770

A recent court filing has exposed the internal workings of Midjourney, revealing that the AI developer openly admitted to scraping the styles of thousands of artists to train its image generation engine. The CEO explicitly acknowledged incorporating specific artist names into the model’s dictionary to allow users to emulate their work through prompts. This disclosure serves as tangible evidence for the ongoing lawsuit filed by artists, who argue that generative AI functions not as innovative technology, but as a mechanism for copyright infringement and labor exploitation. The list includes prominent figures from the science fiction and fantasy communities, such as Hugo Award winners, demonstrating the widespread reliance on professional artistic labor without compensation or permission. By publicly sharing this data, Midjourney inadvertently provided plaintiffs with direct proof that the platform’s capabilities are derived from unauthorized use of copyrighted material. This challenges the company’s narrative of technological novelty and highlights the ethical and legal complexities surrounding the collection of creative data for commercial AI development. This revelation is critical to the open data movement, as it underscores the opaque and non-consensual nature of current training datasets. It illustrates how major technology companies often exploit creative works without transparency, raising urgent questions about intellectual property rights in the age of machine learning. The incident reinforces the need for clear standards regarding data sourcing, consent, and attribution, urging the community to advocate for ethical AI practices that respect the contributions of human creators rather than treating their work as free resources for algorithmic improvement.

Source: file770.com
Published on 2024-01-03