Apple Gets Sued the Same Day Anthropic Settles - September 5, 2025

Authors Grady Hendrix and Jennifer Roberson sued Apple on Friday for using pirated books to train its OpenELM models. The timing wasn't coincidental - it was the exact same day Anthropic agreed to pay $1.5 billion to settle a nearly identical lawsuit. Lawyers know an opportunity when they see one.

Apple getting caught using the same pirated book datasets is especially embarrassing because they've spent years lecturing everyone else about privacy and data ethics. Apple Intelligence marketing emphasizes "user consent" and "on-device processing" - but apparently that ethical stance didn't extend to acquiring training data legally.

The lawsuit specifically targets Apple's OpenELM models, which used "a known body of pirated books" for training. So while Apple was marketing privacy as a premium feature, they were torrenting books just like every other AI company.

Sue All the Things

Friday was apparently "fuck AI companies day" for copyright lawyers. Besides Apple, Warner Bros. also sued Midjourney for training on copyrighted images. The message is clear: Anthropic's $1.5B settlement proved these lawsuits can pay, so every content creator with lawyers is jumping in.

Authors have already sued OpenAI, Microsoft, Meta, and Google. Now entertainment companies are getting in on the action. This isn't random litigation - it's coordinated strategy to establish that AI companies need to pay for training data.

The Judge Alsup ruling in the Anthropic case created the perfect legal framework: using legitimately acquired books for AI training is fair use, but using pirated books is just copyright infringement. This makes the lawsuits much easier to win.

Apple's "Privacy" Hypocrisy

Apple's whole AI strategy is built on the premise that they're more ethical than competitors. Private Cloud Compute keeps user data on Apple servers instead of third-party clouds. Apple Silicon chips process AI requests on-device for privacy.

But all that privacy theater was for user data, not training data. When it came to actually building the AI models, Apple used the same pirated books as everyone else. They just did their piracy behind closed doors while publicly criticizing competitors.

This is a massive reputational risk for Apple. Their premium pricing depends partly on being "the ethical tech company" versus Meta and Google. If courts rule that Apple used identical training practices, what exactly are customers paying extra for? The privacy marketing falls apart if Apple was pirating content just like everyone else.

The Anthropic settlement just proved that AI companies will pay billions to avoid going to court over pirated training data. Authors and their lawyers figured this out, which is why Apple got sued literally hours after the Anthropic settlement was announced.

The New Legal Playbook

This isn't random litigation - it's a coordinated strategy. The Authors Guild and copyright lawyers have figured out the perfect attack: sue right after another company settles, while everyone's attention is on copyright violations. It creates maximum pressure for quick settlements.

The legal framework from the Anthropic case makes these lawsuits easy to win. Using legitimately purchased books for AI training? That's fair use. Downloading millions of books from torrent sites? That's just copyright infringement with fancy branding.

Apple has usually won legal battles through superior resources and careful legal strategy. But this isn't patent trolling or antitrust - it's basic copyright law. Either you licensed the content or you didn't. Apple's usual legal advantages don't help when you got caught using LibGen.

Training Data Just Got Expensive for Everyone

OpenAI already has licensing deals with news organizations like AP and Financial Times. Google negotiated partnerships with publishers. Smart companies saw this coming and started paying for clean data.

Apple has $133 billion in cash, so they can afford systematic licensing deals. But their obsessive secrecy makes negotiations harder. Content creators want transparency about how their work gets used. Apple's "trust us, we're Apple" approach doesn't work for licensing deals.

The Real Winners and Losers

If courts keep ruling against AI companies, here's what happens:

• Big Tech companies can afford licensing deals and legal settlements
• Content creators finally get paid for training data instead of getting ripped off
• AI startups get priced out because they can't afford clean training data
• Open-source models die because nobody can afford to license content for free models

Apple can handle billion-dollar settlements and licensing fees. Meta and Google can too. But if you're a YC startup trying to build the next AI unicorn, you're fucked. You can't afford licensing deals and you can't afford lawsuits.

This fundamentally changes AI economics from "download whatever you want for free" to "pay content creators or get sued." That's probably fair, but it means AI innovation gets concentrated among companies rich enough to pay for training data.