AI Policy Wiki
Dashboard

Anderson v. Microsoft (copyright shareholder-derivative suit)

medium confidence · updated 2026-07-26

June 2026 shareholder-derivative complaint against Microsoft's directors and officers alleging breach of fiduciary duty over the company's OpenAI investment and Azure support for training on unlicensed copyrighted works; the second company, after Adobe, hit with the 'copyright shareholder-derivative' theory.

Anderson v. Microsoft Corporation (No. 2:26-cv-02281, W.D. Wash., filed June 30, 2026) is a shareholder-derivative action brought by a Microsoft shareholder, Anderson, against the company's directors and officers, with Microsoft Corporation as a nominal defendant. The complaint alleges that the defendants breached their fiduciary duties by investing in OpenAI and providing Microsoft's Azure cloud infrastructure to enable OpenAI to train models on unlicensed, allegedly infringing copyrighted works, and by approving 2024 and 2025 proxy statements that downplayed the resulting legal exposure (Source: chatgptiseatingtheworld.com; news.bloomberglaw.com). The filing makes Microsoft the second company, after Adobe, targeted under what commentators have called the "copyright shareholder-derivative" theory.

Infobox

FieldValue
Case number2:26-cv-02281 (W.D. Wash.)
CourtU.S. District Court, Western District of Washington
FiledJune 30, 2026
PlaintiffAnderson, a Microsoft shareholder, suing derivatively on behalf of Microsoft Corporation
DefendantsMicrosoft directors and officers; Microsoft Corporation (nominal defendant)
TypeShareholder-derivative action; breach of fiduciary duty
StatusActive (recently filed)

Allegations

The complaint frames the alleged breach around two lines of conduct. First, it asserts that Microsoft's directors and officers exposed the company to copyright liability through its AI strategy — its investment in and partnership with OpenAI and its supply of Azure infrastructure used to train models on unlicensed copyrighted material. Second, it alleges that the defendants approved 2024 and 2025 proxy statements that defeated shareholder proposals raising concerns about copyright infringement and related litigation, and that those proxy statements contained misrepresentations about the datasets Microsoft used (Source: chatgptiseatingtheworld.com).

Among the specific allegations, the complaint states at paragraph 94 that "Microsoft had violated federal copyright laws in the formulation of its AI strategy and its partnership with AI companies like OpenAI." It also references alleged missteps and misstatements tied to Microsoft's AI products, including Copilot (Source: chatgptiseatingtheworld.com). The complaint quotes language from Microsoft's 2024 proxy statement (at page 88) and 2025 proxy statement (at page 82) describing the datasets used, both filed with the SEC.

The suit is an application of a legal theory first used against Adobe, in which shareholders sue a company's executives for breach of fiduciary duty on the ground that the company trained or enabled the training of AI models on unlicensed copyrighted works and misrepresented its data practices. Adobe faced two such actions before the Microsoft filing — SEIU Pension Plan Master Trust v. Narayen and Hirschberger v. Narayen, both in the Northern District of California, centering on the alleged use of the SlimPajama dataset (Source: chatgptiseatingtheworld.com). The theory recasts copyright-infringement exposure as a corporate-governance and securities question, aimed at directors and officers rather than at the company as a direct copyright defendant, and its spread to a second large AI investor is what makes the Microsoft filing notable. By the count of one copyright-litigation tracker, the action is the twelfth AI-related copyright suit against Microsoft and the 123rd against AI companies in the United States (Source: chatgptiseatingtheworld.com).

The derivative suit is distinct from the direct copyright-infringement claims Microsoft already faces as a co-defendant alongside OpenAI, including the consolidated news-publisher actions in the Southern District of New York (see NYT v. Microsoft, OpenAI et al.).

The complaint itself (Anderson v. Microsoft — Shareholder Derivative Complaint (June 2026)) pleads breach of fiduciary duty and violation of Exchange Act §14(a) over a relevant period beginning January 1, 2022, naming Nadella, Hood, Spataro, Jha, Hoffman, Johnston, List, MacGregor, Mason, Peterson, Pritzker, Scharf, Stanton, and Walmsley individually, with Microsoft as nominal defendant. Its theory is disclosure rather than infringement: it quotes Microsoft's representations that it uses data "in a manner consistent with global copyright laws," does "not train on data from domains listed in the Office of the United States Trade Representative Notorious Markets" list, and trains on data obtained "through negotiated arrangements with publishers and copyright owners" — then sets those against the infringement suits beginning with Authors Guild v. OpenAI & Microsoft Corp., No. 1:23-cv-08292 (S.D.N.Y.), in September 2023. The claim therefore does not require proving infringement, only that the assurances were misleading when made and that directors were on notice.

Relationships

  • litigates: Microsoft — directors and officers are the real defendants; the company is a nominal defendant
  • related: OpenAI — the investment and Azure-training conduct at the center of the alleged breach
  • related: AI Copyright — the underlying copyright-infringement exposure the theory is built on
  • related: NYT v. Microsoft, OpenAI et al. — direct copyright claims against Microsoft and OpenAI, distinct from this derivative action
  • instance-of: copyright shareholder-derivative suits (theory first applied to Adobe)

Sources

  • Chat GPT Is Eating the World / "New copyright shareholder derivative suit v. Microsoft, its directors & officers" (2026-07-01)
  • Bloomberg Law / "Microsoft Top Brass Sued by Shareholder Over AI Copyright Claims" (2026-07-01)
  • Complaint, Anderson v. Microsoft Corp., W.D. Wash. No. 2:26-cv-02281 (filed 2026-06-30) — queued for ingest