Apple hit with another class action lawsuit for alleged copyright infringement (www.engadget.com)

Apple has been hit with a new proposed class-action suit alleging it used pirated books to train its AI models. Neuroscience professors Susana Martinez-Conde and Stephen Macknik of SUNY Downstate say Apple trained its Apple Intelligence models using “shadow libraries” and web‑crawling tools that accessed copyrighted books without authorization, including two of their registered works. This follows a separate, similar author suit filed about a month earlier, and comes amid other high‑profile copyright litigation against AI companies (e.g., The New York Times vs. OpenAI). The industry already has a landmark outcome to watch: Anthropic’s recent $1.5 billion settlement with roughly 500,000 authors for similar claims. For the AI/ML community the case underscores growing legal and operational risk around training data provenance. The allegations target common technical practices—large‑scale web crawling and aggregation of third‑party text—so a loss or large settlement could push firms toward stricter licensing, more rigorous content filters, audit trails for training corpora, and alternative strategies like synthetic or opt‑in datasets. It also raises questions about model training pipelines, recordkeeping, and downstream liability for outputs. Developers, researchers, and companies will be watching closely for how courts balance innovation against copyright protections, since rulings could reshape how foundation models are built and deployed.