What happened
John Carreyrou and five other authors initiated individual copyright lawsuits against xAI, Anthropic, Google, OpenAI, Meta, and Perplexity. The plaintiffs allege these six AI companies utilised copyrighted books, specifically pirated copies downloaded from illegal sources, to train their large language models without authorisation, infringing intellectual property rights. This action diverges from prior class-action suits, with authors pursuing individual claims for increased compensation per infringed work, having rejected Anthropic's $1.5 billion class-action settlement. This marks the first instance of xAI being named in such a copyright infringement case related to AI model training.
Why it matters
This legal action introduces a significant operational constraint regarding the provenance and licensing of training data for large language models. It increases the due diligence requirements for data acquisition and management, particularly concerning copyrighted materials and their source legitimacy. The burden falls primarily on legal, compliance, and data engineering teams to ensure that training datasets are free from infringing content, thereby increasing exposure to intellectual property litigation risks and potential statutory damages for each alleged infringement. This highlights a control gap in verifying the lawful acquisition of training data.
