A prominent New York Times reporter has sued Google, xAI, OpenAI and other AI giants over chatbot training practices, alleging unauthorized use of copyrighted books and other works to train large language models (LLMs) that power AI chatbots. The lawsuit, filed in a California federal court on December 22, 2025, marks a significant escalation in global legal scrutiny of AI training data practices by tech giants.
John Carreyrou, investigative journalist and author of Bad Blood, joined five other authors in accusing companies including Elon Musk’s xAI, Alphabet’s Google, OpenAI, Meta Platforms, Anthropic and Perplexity of feeding their copyrighted works into AI training data without permission. The complaint argues that these firms enriched their chatbots at the expense of authors whose work was used without consent.
Unlike earlier lawsuits, the plaintiffs opted out of a class action, arguing that grouping their cases would let tech companies settle for minimal payouts that undervalue individual claims. The complaint criticized a prior $1.5 billion Anthropic settlement for providing tiny fractions of statutory maximum damages per work.
Legal experts say the case comes amid a broader trend of copyright challenges against AI training practices, including movements in India to require royalties for training on creators’ work. AI training data rights are emerging as a global regulatory flashpoint, with governments and courts increasingly questioning how much content AI firms can use without compensation.
In an industry already under pressure, AI development deals and investments continue to make headlines. For instance, major tech funding developments, such as Amazon’s potential multibillion‑dollar investment in OpenAI, could reshape competitive dynamics even as legal battles unfold. A recent report notes Amazon is in talks to invest up to $10 billion in OpenAI, boosting the latter’s financial and operational muscle even amidst litigation challenges.
Opponents of the lawsuit say proving actual copyright infringement in AI training remains legally complex, especially as companies argue that LLMs transform and generalize learning data rather than copying it directly. But advocates for creators see this legal push as a watershed moment for content rights in the AI age.
