When are consent, licensing, attribution, or compensation required for copyrighted and personal material?
Courts began answering in 2025, partially crediting training as transformative learning while rejecting it for pirated corpora. The supply-chain framing maps who owes whom at each stage from scraping to output.
View on the map → · Open in Browse →
The money and the precedents both arrived: final approval of the ~$1.5B Bartz settlement paid authors about $3,100 a title while releasing only past conduct, the first federal appellate test of fair use for AI training reached oral argument, Europe's first major ruling treated memorization in model weights as reproduction — and the UK abandoned its opt-out proposal after creator backlash.
The algorithm is being used to essentially try to put us out of a job
Reports final approval of the Bartz v. Anthropic settlement (~$3,100 per title to over 300,000 authors) and documents why many authors see the compensation as validating training on their work rather than vindicating them.
This settlement only releases Anthropic from liability for past conduct—specifically, its acquisition, retention, and use of the identified pirated works before August 25, 2025.
The best analytical unpacking of what the settlement does and does not resolve: class certification plus statutory damages created the leverage, outputs and future conduct remain open, and a 'shadow library strategy' template now exists for plaintiffs.
The question has legal language in it, and what we have taught this machine now is how to think like a lawyer.
The first federal appellate test of fair use for AI training reached oral argument in June 2026; the panel's questioning on transformativeness and market harm previews the first binding appellate precedent.
We propose to gather further evidence on how copyright laws are impacting the development and deployment of AI
Analyzes the UK government's March 2026 report abandoning its preferred opt-out text-and-data-mining exception after creator backlash — a major national policy reversal on the consent-and-licensing question.
model parameters sit in a space that copyright doctrine has always treated cautiously: they are closer to facts, statistics and functional logic than to expressive form.
The leading scholarly critique of the Munich court's GEMA v. OpenAI ruling — Europe's first major decision treating memorization in model weights as reproduction — arguing it mischaracterizes lossy compression as storage.