According to TechCrunch, a group of authors including high-profile Theranos whistleblower and “Bad Blood” author John Carreyrou has filed a new lawsuit against six major AI companies: Anthropic, Google, OpenAI, Meta, xAI, and Perplexity. The core accusation is that these companies trained their large language models on pirated copies of the plaintiffs’ books. This action follows a prior class action settlement against Anthropic, where a judge ruled training on pirated material was legal but obtaining it was not. That settlement, worth about $1.5 billion, would allow eligible writers to receive roughly $3,000 each. The new plaintiffs argue this deal “seems to serve [the AI companies], not creators,” letting them off too easily for willful infringement that fuels billion-dollar revenues.
The Legal Groundhog Day
Here’s the thing: this feels like a strategic do-over. The first lawsuit got a mixed ruling that basically created a weird loophole. Sure, you can’t steal the books, but if someone else does and you use them… that’s okay? That’s a distinction that probably doesn’t sit well with anyone who creates for a living. So this new suit, with a different set of plaintiffs and a broader target list, seems designed to attack the heart of the issue: the actual use of the copyrighted material in training. They’re calling the previous settlement a “bargain-basement” way to avoid the true cost. It’s a clear signal that a chunk of the creative community isn’t going to accept the first deal offered. They want a precedent that holds the tech giants directly accountable for the data they ingest, not just the shady websites they might get it from.
Winners, Losers, and the Future of AI Fuel
So who wins if the authors succeed? In the short term, it’s a potential windfall for copyright holders and a massive new cost of doing business for AI firms. Training data suddenly gets a lot more expensive. That could slow down the breakneck pace of model development and solidify the advantage of the deepest-pocketed players like Google and Meta. The losers? Smaller AI startups and open-source projects that rely on scraping publicly available data. The entire economics of the field shifts. But look, if the authors lose? Then the current “ask for forgiveness later” model gets a huge endorsement. AI companies would have even less incentive to clean up their data sourcing practices. It creates a perverse system where the value is in the illegal aggregation, not the creation. Basically, we’re deciding right now what raw materials are fair game for the most important tech of the next decade. That’s a pretty big deal.
