Let’s Do Data Science: “U.S. courts remain split on whether training AI models on copyrighted material is fair use, after Judge William Alsup ruled in June 2025 that training on purchased, digitized books was fair use while Judge Vince Chhabria reached the opposite conclusion two days later in a related San Francisco case. The split has taken on new urgency after Google’s president of global affairs, Kent Walker, published a 21-page white paper on June 25, 2026 arguing regulation should target AI outputs rather than training inputs, a stance that clashes with Colorado’s revised AI Act and California’s Training Data Transparency Act, both focused on input disclosure. Publisher group Digital Content Next has sent Common Crawl a cease-and-desist letter rejecting Google’s opt-out framework, arguing copyright law “is not an opt-out regime.” For practitioners, the unresolved legal and regulatory divide keeps dataset provenance and licensing strategy a live risk factor.”

• June 24, 2025 Judge William Alsup rules that training AI models on purchased, digitized copyrighted books can be fair use.
• June 26, 2025 Judge Vince Chabria reaches the opposite conclusion in a related San Francisco case involving AI training.
• January 1, 2026 California’s Training Data Transparency Act takes effect, requiring AI developers to disclose training-data sources.
• June 25, 2026 Google publishes a white paper arguing AI regulation should target outputs rather than training inputs.
• July 1, 2026 PYMNTS reports that Digital Content Next has sent Common Crawl a cease-and-desist letter over web-scraping opt-outs.