Cognella, Cambronne and Chicken Soup for the Soul v. SpaceX Super Intelligence

Who is suing, and over what

These are three separate lawsuits, filed the same day by the same lawyers, and this page treats them together because their allegations about the defendants are nearly word for word the same. What differs is who owns the books.

None of the three is a class action. Each complaint says the plaintiffs chose to sue individually for statutory damages rather than wait for a class settlement (Cognella ¶ 12). That is the same choice the firms made in their suits against Meta (see Book authors v. Meta and Levine v. Meta).

Who the defendants are

The complaints name three related Nevada companies and call them together “xAI” (Cognella ¶ 1). Each is now named SpaceX Super Intelligence and was earlier named X.AI and then SpaceXAI; Holdings is also successor by merger to X.AI Holdings Corp. As the complaints describe them, Holdings owns the Corp., which owns the LLC. On information and belief, they allege that the Corp. built and trained Grok, that the LLC develops and distributes it, and that Holdings directs, finances and profits from it (Cognella ¶¶ 17-22; Cambronne ¶¶ 21-26). The complaints place the alleged conduct both before and after SpaceX’s acquisition of xAI in February 2026. X Corp., the company that runs the X platform, is not a defendant.

Four counts

Paragraph numbers shift slightly between the three complaints. Chicken Soup numbers its counts about two paragraphs lower than Cognella.

How the books are said to have been obtained

Shadow libraries. The complaints say Grok was trained on Books3 (part of The Pile), Library Genesis, Z-Library and Anna’s Archive, and that the defendants downloaded them by BitTorrent (Cognella ¶¶ 50-62, 74-75). Part of the support offered is Grok itself: the complaints describe answers in which the model discussed Library Genesis as a training source (Cognella ¶¶ 67-70). For Chicken Soup’s titles, presence in those collections is pleaded by inference rather than by title (Chicken Soup ¶¶ 15, 30).

Scanned print books. Count II rests on allegations that the defendants bought print books and scanned them for training, citing a public post by Elon Musk in July 2026, and that copyright pages were removed in the process (Cognella ¶¶ 6-7, 81-82, 87-99). The complaints tie this to the later Grok models.

Outputs. There is no separate count for model outputs. The complaints instead cite an academic study reporting that a Grok model reproduced much of a well-known novel without a jailbreak (Cognella ¶¶ 72, 103; see the paper listed under Documents), and argue that Grok competes with and dilutes the market for the works, citing the Copyright Office’s dilution analysis and Judge Chhabria’s opinion in Kadrey v. Meta (Cognella ¶ 102).

The scanning count and Bartz

The complaints cite Bartz v. Anthropic for its facts about pirated libraries and print scanning (Cognella ¶ 74 n.100, ¶ 88 n.116). They do not adopt its fair-use reasoning. As ChatGPT Is Eating the World notes in its 5 November post, Judge Alsup’s Bartz ruling went the other way on scanning purchased print books. Count II asks a different judge to reach a different answer, as the same firms’ Meta complaints do.

Where these sit

ChatGPT Is Eating the World counts these three as suits 162 to 164 against AI companies in U.S. courts, and reports that Stris & Maher has now filed more AI copyright suits than any other plaintiffs’ firm. With its five suits against Meta and these three, the firm is pressing the same two theories against Meta and xAI: that downloading books from shadow libraries is itself infringement, and that scanning print books for training is not fair use.