Julia Angwin’s Complaint Pictures Two Notes Under Her Name.

Her complaint says the notes appeared “in Grammarly’s Expert Review” and pleads lost control over her name’s commercial use.

Page 5 of Julia Angwin’s complaint pictures two margin notes, each headed “Julia Angwin.” One reads “Clarify crowd-sourced method upfront.” The other reads “Simplify technical tracker description.” The complaint calls them “examples of comments attributed to Julia Angwin in Grammarly’s Expert Review.” It was filed March 11, 2026 as Angwin v. Superhuman Platform, Inc., No. 1:26-cv-02005 (S.D.N.Y.), and sues the defendant over that tool. Beside the first, the picture highlights a passage beginning “The Markup also found the Meta Pixel installed inside the password-protected patient portals of seven health systems.”

The complaint says Angwin learned that Grammarly “had appropriated her name and identity” when she read a report by another journalist, cited as Casey Newton’s Platformer column of March 9, 2026. In a newsletter post on March 14, 2026, she wrote that she “was both honored and appalled to be included in a lineup of experts that included the bestselling novelist Stephen King and the feminist author bell hooks.” Both accounts are hers. They are not two confirmations of how she learned of it.

What the complaint says the tool did

Paragraph 17 alleges that the tool tells users it is “reading your text” and “finding experts to review your piece,” then that it is “Applying ideas from” experts. Its example is “Applying ideas from Julia Angwin,” and the complaint says the product described her as an investigative journalist and Pulitzer Prize finalist. Paragraph 19 alleges that when a user clicks a comment, Grammarly states that the “suggestion is inspired” by the expert, and that the user can choose to insert an example of what the expert suggests by clicking a button.

The complaint adds that Grammarly “is analyzing Ms. Angwin’s work and, based on that work, is drawing conclusions about what advice she would give.” It also alleges that Grammarly “failed to obtain the consent of Ms. Angwin and hundreds of other ‘experts’” before it launched the product nationwide. All of this is the complaint’s account of the tool. Neither the product nor the screenshot filed as Dkt 23 Ex. A has been examined here.

What the complaint pleads

The complaint pleads four claims. The first is a California common-law claim for right of publicity. Two rest on statutes, California Civil Code section 3344 and New York Civil Rights Law sections 50 and 51. The fourth, unjust enrichment, is pleaded in the alternative. For the first, the complaint states that the claim “is established by” four elements: “(1) the defendant’s use of the plaintiff’s identity; (2) appropriation of the plaintiff’s name or likeness to the defendant’s advantage; (3) a lack of consent; and (4) injury to the plaintiff.”

Paragraph 53 pleads that “Plaintiff and the Class Members have been injured by the misappropriation of their names and identities in Expert Review, including by violating their privacy.” It pleads three more. One is “preventing them from retaining control over the commercial use of their identities.” Another is “associating work product with the ‘Experts’ even though the ‘Experts’ had no knowledge or control over the quality of the work.” The third is “causing them economic injury by not compensating them for the use of their names.” That is four injuries in one paragraph, and one of them is economic. Paragraphs 24 and 33 name Angwin herself. Paragraph 24 says she was concerned that the practice “prevented her from controlling the commercial use of her name and identity” and the work the public associates with them. Paragraph 33 alleges that Superhuman’s conduct “has caused Ms. Angwin and other journalists, authors, and editors to lose control of the commercial use of their names and identities” and of the work the public associates with them.

Whose advice it was

The complaint’s paragraph 3 says one aim is “to stop Grammarly from attributing words to them that they never uttered and advice that they never gave.” Paragraph 22 uses the words “writing advice that these experts did not give.” The same frame recurs in at least four later paragraphs. As filed on March 11, 2026, the complaint does not specifically allege that either pictured note is false, fabricated, incorrect or inaccurate. The complaint’s language about wrong or disputed advice is general. Paragraph 22 says Grammarly “may draw incorrect conclusions about her writing,” and paragraph 25 imagines a user who “did not agree with her writing advice.”

On our reading, the complaint as filed frames its case around her name. The complaint alleges words and advice attributed to her that she never gave, pleads four injuries, and does not specifically allege that either pictured note was false. The elements it lists for its California common-law claim do not include accuracy. That is our reading, not a statement of what the law requires, and not a claim that the harm it pleads is only about the name. We describe the two pictured notes by form only, as short imperative notes about clarifying and simplifying; that describes those two notes and no more. Her March 14, 2026 post says a friend showed her screenshots and videos of suggestions using her name, and that the edits “ranged from anodyne comments” to “outright fabrications.” The joint letter of June 23, 2026, signed by counsel for both sides, says she “intends to amend the complaint” and proposes, if mediation is unsuccessful, an amended complaint by November 16, 2026.

What Superhuman has filed, and what the docket lists

The complaint alleges that in August 2025 Grammarly launched Expert Review “for its paying customers.” Superhuman’s brief of June 15, 2026 says Expert Review “was available to free and subscription-based Grammarly user accounts between August 2025 and March 2026 and is currently disabled,” citing paragraph 4 of the Kepczynski declaration (Dkt 19). This edition does not reconcile the two. The same brief calls the feature an agent embedded in the Grammarly product, in support of its motion to compel arbitration.

The brief says the plaintiff’s “claims hinge on how the Expert Review feature operated,” including “what large language model (LLM) was used for the prompts, what prompts were issued, and whether consent was ever obtained.” That is the brief’s argument about arbitration, not an admission. As of October 4, 2026, none of the sources this edition read names the model or the prompts.

The docket, re-pulled October 4, 2026, lists a joint letter motion of June 23, 2026 (Dkt 24) to stay the litigation “for the Parties to Pursue Mediation.” Judge John G. Koeltl’s order of June 24, 2026 stays the case until October 15, 2026 and directs the parties to give the court “a status report on or before October 15, 2026.” As of the CourtListener docket re-pulled October 4, 2026, the docket lists no order resolving the motion to compel arbitration (Dkt 17) or the motion to dismiss (Dkt 21), and no amended complaint.

The Agentics read

Agentics reads this record through Digital Derangement Syndrome™ (DDS™): AI-mediated systems failing to recognize legitimate authority. The closest fit, and only a fit, is Trust Transfer Failure, the fourth clinical characteristic: expertise that exists, in full, in the physical world and has never been translated into machine-readable trust. Here the direction is reversed. If the complaint is right, her name was put to work on advice she never gave.

Where the argument stops

This edition read the complaint, Superhuman’s brief (Dkt 18), the docket, Dkts 24 to 26, Angwin’s post, and TechCrunch and TechRadar article text. Not read: Dkts 16, 17 and 19 to 23, including the Kepczynski declaration, Dkt 22 and the Dkt 23 Ex. A screenshot; Platformer, The Atlantic, the Verge Decoder transcript and Mehrotra’s LinkedIn post. The docket is CourtListener’s, sourced from PACER and RECAP, and may not be up to date. Every contested fact here is the complaint’s allegation, Angwin’s account or Superhuman’s litigation position.

The parties’ letter records her intent to amend, and this edition does not say what an amendment will plead. The Superhuman filing examined here is its June 15, 2026 brief (Dkt 18). This edition takes no side, and does not say whether any claim will succeed. The mapping to DDS™ is our editorial reading, not a finding by any court or party.