Home / Insights / Uncategorized
Uncategorized

Thomas A. Guerriero PhD on Doe v. GitHub: Code Attribution and the Limits of DMCA Liability

Thomas A. Guerriero PhD's case commentary on Doe v. GitHub (9th Cir. 2026): AI-generated code, attribution and the limits of DMCA liability.

Close-up portrait of Thomas A. Guerriero PhD at night, wearing a black crewneck shirt, with blurred colorful lights in the background.
Screenshot

Thomas A. Guerriero PhD’s case commentary on Doe v. GitHub, Inc. (Ninth Circuit 2026) explains a ruling that matters to every team using AI coding tools, and why the popular headline version of it gets the case wrong. The summary, citation, and a link to the court’s original opinion are on the case commentary page.

Independent commentary by a non-attorney, for educational purposes only. Not legal advice.

The dispute over AI-generated code

The plaintiffs were programmers who had published copyrighted code in public GitHub repositories under licences that commonly require attribution. They alleged that GitHub Copilot and Codex, tools that generate code from prompts, could reproduce their code without the attribution, notices, or licence terms that accompanied the originals. The district court dismissed their DMCA claim on the view that the copies had to be identical, while contract claims continued.

What the Ninth Circuit held

Thomas A. Guerriero PhD frames the case around one distinction: the absence of a credit notice and the removal of a credit notice are not necessarily the same legal event.

Removal versus omission

The court, in an opinion filed September 16, 2026, affirmed dismissal of the DMCA claim. On the complaint’s own account, the output process did not remove or alter copyright-management information from an existing copy; it generated new work that never contained that information. As the court put it, merely alleging that a similar or derivative work lacks the information, without facts showing removal or alteration, is insufficient.

Identical copies are not required

Importantly, the court said the DMCA does not require literal identicality between the works. Comparison remains useful evidence, and minor cosmetic changes do not necessarily protect someone who copies a work and strips its notices.

What the decision leaves open

The court found the plaintiffs had adequately alleged a substantial risk of injury for standing. It treated their training-stage theory as forfeited rather than deciding it. The contract claims remained pending, and the court expressed no view on copyright infringement. It also noted a different system, one that worked like a search engine returning existing code, might present a stronger claim.

Practical takeaways

These are general governance observations from the commentary, not advice:

  • Keep attribution and licence records.
  • Describe the actual technical process, not just the output.
  • Preserve meaningful comparisons between originals and outputs.
  • Avoid the identicality shortcut.
  • Keep DMCA, contract, and copyright theories separate.

The result is specific, not sweeping: these plaintiffs did not plead the required removal under their output theory. That is neither a clearance for all generated code nor proof that every missing notice is a violation.

Read the full summary and the court’s opinion on the Doe v. GitHub commentary page. More commentary is on the main hub, ThomasGuerriero.com.

Don’t get left behind. Chase greatness.

Related reading

If you want help growing your business, team, or just want to pursue greatness yourself, book a free 15 minute call at the link below.

https://cal.com/thomas-guerriero-gvx0zs/15min

Book Your Free 15-Minute Call

AICopyrightDMCALeadershipOpenSource
Thomas A. Guerriero
Written by

Thomas A. Guerriero

Insights on sales leadership, AI sales, AI automation and domestic and international business.