March 20, 2026
Settle for freedom, not cash
FSF Threatens Anthropic over Infringed Copyright: Share Your LLMs Freel
FSF says keep your cash, free the AI—commenters cheer, scoff, and meme
TLDR: FSF replied to an AI copyright settlement notice by saying their works are already free and they’d prefer open models, data, and code over money. Commenters split between calling it noble, performative, or impractical—sparking memes, a car‑theft analogy, and a compromise idea to open models after a delay.
Drama alert: the Free Software Foundation (FSF) got a notice about a class‑action settlement against Anthropic over training its AI on books scraped from pirate libraries. A judge said using books to train was fair use, but the legality of downloading them was murky—so the parties settled and started offering payouts. The FSF’s response? Their featured book is under a free license, so they don’t want money; they want freedom—meaning: release the model, the training data, the settings, and the code to users. They even say that if they ever joined a case like this, they’d ask for openness as compensation.
Cue comment section fireworks. One camp yelled “clickbait,” with mjg59 asking, “Where’s the threat?” Meanwhile, politelemon turned it into a meme: “We don’t usually sue… we settle for freedom.” Supporters cheered the FSF’s posture as the only honest response to AI trained on internet content. mcv threw shade at Anthropic: if your business model depends on using others’ works without permission, maybe you don’t have a business model. The angriest take came from slopinthebag, comparing excuses for AI firms to letting a car thief off because they “need to get to work.” A middle path surfaced from kouteiheika: release models for free after X months. Verdict from the crowd: bold idealism vs. corporate pragmatism—plus a whole lot of memes about “unlocking the AI” and “settling for freedom.”
Key Points
- •FSF received a settlement notice related to the class action Bartz v. Anthropic over alleged downloading of works for LLM training.
- •A district court ruled that using books to train LLMs was fair use, while leaving the legality of downloading those works unresolved for trial.
- •Parties chose to settle before trial and are contacting potential rightsholders with payment offers in lieu of potential damages.
- •FSF holds copyrights to GNU programs and books published under free licenses, including a GNU FDL-licensed book found in Anthropic’s training datasets.
- •FSF urges Anthropic and similar developers to release training inputs, models, configurations, and source code to users, and would seek such remedies if its rights were violated.