Gather — 9 July 2026

Courts Split on Fair Use; Bartz v. Anthropic Settlement #

• Courts Split on Whether AI Training Is Fair Use (PYMNTS) — Bartz v. Anthropic: court ruled training on copyrighted books constitutes fair use, but storing pirated copies does not. $1.5B settlement, ~$3,000 per work. A separate court (Meta case, June 2025) also found training transformative/fair use but disagreed on sourcing — holding training = fair use regardless of whether materials were legitimately obtained. Courts agreeing on the outcome but diverging on the reasoning creates an unstable precedent.
• AI in Litigation: An Update on AI Copyright Cases in 2026 (Norton Rose Fulbright) — July 2026 tracker update. Key cases in active litigation: Udio DMCA stream-ripping claims (denial of motion to dismiss May 21 — those claims proceed); July federal hearing on music training data could force AI companies to pay for training data; AI Copyright Lawsuit Escalates: Hagens Berman joins Suno/Udio fight with tobacco-deal firepower.

July Federal Hearing: Music Training Data #

• AI Copyright Lawsuit Escalates: Firm Behind $260B Tobacco Deal Joins Suno and Udio Fight (TechTimes) — Hagens Berman (the firm behind the $260B tobacco settlement) joins the Suno/Udio music copyright fight. DMCA stream-ripping claims survive motion to dismiss — this case now has discovery teeth and a well-resourced plaintiff firm.

Copyright.gov Part 3 Report #

• Copyright and Artificial Intelligence — Part 3: Generative AI Training (U.S. Copyright Office) — Pre-publication version. The Copyright Office’s formal position on AI training data — a direct input to any legislative proposals. The analysis will shape whether Congress moves on safe harbour modifications or licensing frameworks.

CJEU Pending: September 3 AG Opinion #

• Like Company v Google: Court of Justice Holds First-Ever Hearing on Generative AI and Copyright (Bird & Bird) — No new developments since the March 10 hearing. Advocate General opinion still expected September 3, 2026. Three questions: (1) does LLM training constitute reproduction?; (2) does the TDM exception cover commercial LLM training?; (3) does AI-generated output reproducing protected content constitute infringement attributable to the AI provider?

Cross-links #

• Open vs Closed AI Ecosystems Courts-split-on-sourcing finding is particularly relevant for open-weight models trained on scraped data — the Meta ruling’s “fair use regardless of sourcing” position would benefit open-weight developers if it stands.
• AI Impact on Society The $1.5B Bartz settlement establishes a per-work floor ($3,000) that will inform licensing negotiations and future cases.

Meta-observations #

• Emerging theme: Courts are splitting not on the outcome (training = fair use, broadly) but on the reasoning (does sourcing matter?). The Meta/Bartz divergence on sourcing is the key unsettled question heading into the September CJEU opinion.
• Quality signal: Norton Rose Fulbright’s AI litigation tracker is the most comprehensive single-source update on active cases.
• Keyword suggestion: "Bartz Anthropic" settlement OR "fair use training data sourcing" 2026 to track the sourcing-question divergence.