Key Points
- Writers, led by John Carreyrou, sue Anthropic, Google, OpenAI, Meta, xAI and Perplexity.
- Plaintiffs allege the AI firms trained models on pirated copies of their books.
- A prior Anthropic settlement of $1.5 billion gave eligible writers about $3,000 each.
- Authors claim the settlement favors AI companies and fails to hold them accountable.
- The lawsuit seeks to prevent “bargain‑basement” payouts that extinguish high‑value claims.
- Outcome could affect how AI companies acquire and use copyrighted training data.
Background
A group of writers, including John Carreyrou—best known for exposing Theranos and authoring the bestseller “Bad Blood”—has launched a new lawsuit targeting six leading AI companies: Anthropic, Google, OpenAI, Meta, xAI and Perplexity. The plaintiffs assert that these firms incorporated pirated copies of their books into the training data for large language models, thereby infringing the authors’ copyrights.
Earlier Legal Context
The lawsuit references a prior class‑action suit against Anthropic that reached a settlement of $1.5 billion. In that case, a judge ruled that while it is legal for AI firms to train on pirated books, the act of pirating the books themselves remains illegal. Under the settlement, eligible writers receive roughly $3,000 each, a figure many authors consider insufficient given the revenue generated by the AI models.
New Lawsuit Claims
The current complaint argues that the Anthropic settlement and similar agreements serve the interests of AI companies rather than the creators whose works were used without permission. The plaintiffs contend that the settlement amounts are “bargain‑basement rates” that extinguish thousands of high‑value claims, effectively allowing the AI firms to avoid full responsibility for the alleged infringement.
Authors’ Concerns
Authors involved in the lawsuit express dissatisfaction with the existing settlement framework, stating that it does not hold AI companies accountable for the commercial benefits derived from their copyrighted material. They emphasize that the settlement appears to prioritize the financial interests of the AI firms, leaving creators with minimal compensation for the extensive use of their intellectual property in training data.
Potential Impact
If successful, the lawsuit could reshape how AI companies source training data and enforce copyright protections. It may also prompt a reassessment of settlement structures to ensure that creators receive compensation commensurate with the value generated by AI models that rely on their works.
Source: techcrunch.com