Hollywood studios sue Chinese AI company for alleged copyright infringement (www.ft.com)

Major Hollywood studios have filed a lawsuit against a Chinese AI company, alleging the firm trained generative models on copyrighted films, scripts and related audiovisual material without authorization and is using those models to produce derivative content. The complaint reportedly seeks damages and injunctive relief and asks the court to force disclosure of the company’s training data and model workflows. Studios argue the models reproduce protected creative elements and thus infringe copyright, while the defendant is likely to counter that use for training is transformative or fall under other defenses — setting up a test case for how copyright law applies to large-scale model training and multimedia synthesis. For the AI/ML community this raises immediate technical and operational stakes: provenance of training data, dataset curation, and licensing practices are now central legal risks rather than academic concerns. Expect pressure to document data sources (datasheets), restrict ingestion of copyrighted audiovisual works, employ filters or opt-out mechanisms, and adopt watermarking or provenance metadata for synthetic outputs. The case could compel disclosure of datasets and model weights, influence cross-border enforcement norms, and chill or complicate open-source model sharing and multimodal research unless clearer licensing frameworks and technical mitigations for copyright-sensitive content are adopted.