Jump to content

News-Anthropic-1.5B-Settlement-2026

From AI Law Wiki

July 20, 2026 update — U.S. District Judge Araceli Martínez-Olguín granted final approval to Anthropic's $1.5 billion settlement in Bartz et al. v. Anthropic PBC, resolving the certified class claims over books acquired from pirate libraries for Claude training.[1][2][3]

Background

[edit]

The lead plaintiffs are authors Andrea Bartz, Kirk Wallace Johnson, and Charles Graeber, who alleged that Anthropic's Claude AI assistant was trained on pirated copies of books from "shadow library" websites LibGen and PiLiMi.[4][5] The settlement followed partial summary judgment favoring plaintiffs on pirated books, finding that such use was not fair use.[4]

Details

[edit]

The $1.5 billion fund covers approximately 500,000 works at roughly $3,000 per work, with additional payments if more works are added.[4][5] Payments are structured in installments: $300M by October 2, 2025; $300M post-final approval; $450M by September 25, 2026; and $450M by September 25, 2027.[4][6] The settlement covers only past ingestion claims up to August 25, 2025; it does not release claims for future training, AI outputs, or commercial model use from these datasets, and includes dataset destruction provisions.[4][7]

Timeline

[edit]
  • Preliminary approval: September 25, 2025 (Judge William Alsup)[5]
  • Claimants deadline: March 30, 2026 (440,490 of 482,460 eligible works claimed — 91.3% claim rate)[8][9]
  • Opt-outs: <0.5% of Works List; 41 objections filed[9]
  • Objections unsealed: April 2026 (Dkt. Nos. 544, 596, 598, 600, 601, 602, and subsequently Dkt. 630, 640, 641)[10]
  • Judge re-assignment: Case randomly reassigned to Judge Araceli Martínez-Olguín following Judge Alsup's move to inactive status[8]
  • Final approval hearing: Held May 14, 2026; Judge Martínez-Olguín granted final approval July 20, 2026.[9][1][2]
  • Case terminated: July 20, 2026.[3]

Objections

[edit]

Class objections include:[10][11]

  • Publisher favoritism: Objector Professor Bishop (Dkt. 630) argues the settlement systematically favors publishers over authors — publishers could claim roughly 50% or more of the total settlement through royalty presumptions, despite the suit being author-led[10]
  • Foreign work exclusion: Bishop also challenges the exclusion of over 2 million foreign and non-U.S.-registered works, arguing this improperly narrows the class[10]
  • Late notice: Objector Esquivel received notice of the settlement on approximately March 3, 2026 — nearly a month after the February 9 opt-out deadline had expired[10]
  • Group registration undercounting (Dkt. 641): Multiple books registered under a single copyright registration are treated as one "claimable work," dramatically undercompensating prolific authors (e.g., 40+ books counted as a single work)[10]
  • Dangerous precedent (Dkt. 640): The settlement allows AI companies to settle mass piracy at a discounted rate — cheaper than licensing — setting a harmful precedent for future AI copyright disputes[10]
  • Class counsel conflicts: References Judge Alsup's December 2025 concerns about undisclosed fee-sharing arrangements between class counsel[11]
  • Inadequate compensation: Base awards of approximately $3,000 per work before deductions for attorneys' fees and costs, with a tiered system paying more for "important" (nonfiction) books over fiction[11]

The May 14 fairness hearing preceded the July 20 final-approval order; payments remain subject to appeals.[12][2]

Final approval

[edit]

Judge Martínez-Olguín granted final approval on July 20, 2026, and the federal docket records the case as terminated that day.[1][2][3] The settlement fund covers more than 480,000 books at roughly $3,000 per work, requires destruction of files obtained from Library Genesis and Pirate Library Mirror, and leaves authors free to pursue claims over future training or model outputs not released by the deal.[2] More than 90 percent of eligible works drew claims, while opt-outs may continue separate litigation and distributions remain subject to appeals.[2]

Broader context

[edit]

Anthropic was the first major AI company to settle one of the foundational AI training copyright cases. Class counsel seeks approximately $319M in fees and costs (21.26% of the fund), with named plaintiffs requesting $50K each.[4][13]

[edit]

References

[edit]
  1. ↑ 1.0 1.1 1.2 Reuters, "US judge approves Anthropic's $1.5 billion settlement of copyright lawsuit" (July 20, 2026)
  2. ↑ 2.0 2.1 2.2 2.3 2.4 2.5 Unite.AI, "Anthropic’s Book-Piracy Settlement Wins Final Court Approval" (July 20, 2026)
  3. ↑ 3.0 3.1 3.2 CourtListener docket, Bartz v. Anthropic PBC, No. 4:24-cv-05417 (N.D. Cal.)
  4. ↑ 4.0 4.1 4.2 4.3 4.4 4.5 Cite error: Invalid <ref> tag; no text was provided for refs named susman
  5. ↑ 5.0 5.1 5.2 Cite error: Invalid <ref> tag; no text was provided for refs named authorsguild
  6. ↑ Cite error: Invalid <ref> tag; no text was provided for refs named copyrightalliance
  7. ↑ Template:Cite web
  8. ↑ 8.0 8.1 Template:Cite web
  9. ↑ 9.0 9.1 9.2 Template:Cite web
  10. ↑ 10.0 10.1 10.2 10.3 10.4 10.5 10.6 Template:Cite web
  11. ↑ 11.0 11.1 11.2 Template:Cite web
  12. ↑ Template:Cite web
  13. ↑ Template:Cite web
Retrieved from "https://ailawwiki.com/index.php?title=News-Anthropic-1.5B-Settlement-2026&oldid=2080"