AI Copyright Lawsuits Infographic

What this graphic shows

The AI Docket Wall puts every U.S. copyright lawsuit against an AI company in one public tracker's list on a single chart. Each of the 160 cells in Figure 1 is one docket entry, placed in a row for the company named first. Blue cells are suits brought by authors and book rights holders; the legend lets you highlight music, news or any other group instead. Outlined cells are suits that have settled, been dismissed or gone to appeal. Almost every cell is solid.

The other cards show who is being sued, who is suing, where else in the world these cases are being filed, and what the first large settlement, Bartz v. Anthropic, actually paid.

OpenAI is the most-sued AI company

OpenAI is named as a defendant in 30 of the 160 tracked suits, almost one in five. Meta (21), Microsoft (19) and Anthropic (15) follow. Much of Microsoft's count comes from its partnership with OpenAI: 13 of the 19 suits naming Microsoft also name OpenAI. The music generator Suno is fifth with 9.

Authors lead the plaintiffs

Authors and book rights holders brought 61 of the 160 suits, more than a third. Music rights holders brought 28 and news and media companies 24. Most of these suits claim that copyrighted work was copied to train AI models without permission or payment.

The fight is not only American. The tracker's international list has 37 more cases. Canada has the most, 14, and nine of those were brought by authors.

The first big payout: Bartz v. Anthropic

In June 2025, Judge William Alsup ruled that using books to train Anthropic's Claude models was "exceedingly transformative" and a fair use. He refused to extend that protection to the central library Anthropic built from pirated copies. The court found the company had downloaded over seven million pirated copies of books, at least five million from Library Genesis and at least two million from the Pirate Library Mirror, and ordered a trial on those copies.

Anthropic settled instead, for $1.5 billion, which the court called "the largest copyright class action settlement in history." The court granted final approval on July 20, 2026. The class covers 482,460 works, about $3,000 each before fees and costs, which is four times the $750 statutory minimum. By April 16, 2026, claims had been filed for 440,490 works, or 91.3% of the list. Class counsel asked for $187.5 million in fees (12.5%); the court awarded $101,561,111. The settlement administrator expects to send the first payments on or before November 15, 2026.

Most of the law is still unwritten

Only 10 of the 160 tracked U.S. suits are marked settled, dismissed or on appeal, and trial courts have ruled on fair use in just three. Two days after the Bartz ruling, Judge Vince Chhabria granted Meta summary judgment in Kadrey v. Meta, but wrote that the ruling "does not stand for the proposition that Meta's use of copyrighted materials to train its language models is lawful," only that the thirteen authors in that case had made the wrong arguments. In February 2025, in Thomson Reuters v. Ross Intelligence, Judge Stephanos Bibas rejected a fair use defense for copying 2,243 Westlaw headnotes to train an AI legal search tool. He noted that "only non-generative AI is before me today." That case is now on appeal.

Methodology & caveats

Case data comes from the ChatGPT Is Eating the World AI Copyright Case Tracker, U.S. list updated October 30, 2026. Its headline count is 164 U.S. cases; its list has 160 docket entries because some entries cover consolidated cases. All counts in Figures 1 to 4 and the 94% figure are data.co tallies of that list, made on November 6, 2026, using the tracker's own lead defendant, plaintiff type and status for each case. We merged entity names (Meta Platforms into Meta, the OpenAI entities into OpenAI) and, in Figure 1, GitHub into Microsoft. One status was updated: the tracker lists Bartz v. Anthropic as active pending court approval, but the court granted final approval on July 20, 2026, so we show it as settled. The tracker covers copyright and related claims against AI developers. It may not include every filing, and its international list is a selection.

Bartz figures come from the court's orders and the court-appointed settlement administrator. The roughly $3,000 per work is the court's own estimate before fees and costs. The 25% fee cap in Figure 7 is derived from the settlement agreement's percentage limit applied to the $1.5 billion fund. The Kadrey and Ross opinions are cited from copies hosted by the tracker. Statutory damage ranges are from 17 U.S.C. § 504(c) as published by the U.S. Copyright Office. Nothing here is legal advice.