"AI's Visual Plagiarism Problem: Navigating Copyright Law in the Age of Generative Models"

OpenAI claims it's "impossible" to train top-tier AI models without using copyrighted material, stating that public domain data would result in sub-par software. A recent IEEE report found that OpenAI's DALL-E 3 and Midjourney's AI services can recreate copyrighted scenes, leading to allegations of copyright infringement. The issue of liability for AI-generated infringing content remains contentious, with lawsuits from entities like the New York Times and book authors. OpenAI defends its practices, arguing that training on copyrighted material is essential for AI models to function effectively, while critics question the ethical and legal implications of using copyrighted content for training.