U.S. District Court for the District of Columbia, Judge Amit P. Mehta (2025), U.S. District Court for the District of Columbia — https://storage.courtlistener.com/recap/gov.uscourts.dcd.223205/gov.uscourts.dcd.223205.1436.0_1.pdf
Key takeaways
- Having found Google liable under Sherman Act §2 in 2024, the court imposed a remedy that bars exclusive distribution contracts covering Search, Chrome, Assistant and the Gemini app and requires some search-index and user-interaction data sharing and syndication to “Qualified Competitors”. It declined to order a Chrome or Android divestiture, a payment ban or choice screens, and it set a six-year term.
Seeded from the abstract and a skim, before the reading in NOTE-004. What follows is what the work says about itself.
This is the remedies-phase memorandum opinion that followed the court’s 2024 liability ruling that Google unlawfully maintained monopolies in general search and search text ads through exclusive default-distribution agreements. The court says generative AI changed the remedies proceedings: GenAI chatbots are a nascent competitive threat, and part of the court’s aim is to keep Google’s search dominance from carrying over into GenAI. The court accepts Google’s proposed remedies with modifications and adopts some of the plaintiffs’ proposals. It rejects structural divestitures and several behavioral remedies as overreaching or unconnected to the violation. It orders the parties to submit a revised final judgment by September 10, 2025.
Standing in the record
Filed by the reading-time triage of 2026-09-25: 1,040 seconds of active reading over 4 sessions in the papers-feed tracker. It was filed Deferred, unread. NOTE-004 is the close reading of 2026-09-25, and it placed the work: Active — a primary legal source, and the most detailed judicial record to date of how LLMs, grounding/RAG and the GenAI market were described under oath. It is worth reading for that and for the design of its data-sharing remedy. Its technical content is lay-level, built from testimony, and carries no instruction for ML practice.
When seeded, before the reading in NOTE-004, the entry gave a priority for a deeper reading: high — the owner spent substantial time on it across 4 captures (t=1040 s, n=4), and its GenAI findings of fact are load-bearing for how courts understand LLMs and search.
What the seeding asked a deeper reading to check (NOTE-004 is that reading):
- It is one of the first major judicial treatments of LLMs and GenAI as a competitive substitute for search. The findings of fact on grounding and on the role of search indexes for LLMs (pp. 28-36) are directly relevant to ML practice.
- The data-sharing and syndication remedies (Qualified Competitors, privacy-enhancing anonymization of user-side data) should be checked in the later final judgment for exact scope.
- The appeal status and any later modifications are unverified here.
Access when seeded: I retrieved the full 230-page PDF from the CourtListener RECAP archive. The page header reads “Case 1:20-cv-03010-APM Document 1436 Filed 09/02/25”. The caption covers the consolidated cases United States et al. v. Google LLC (No. 20-cv-3010) and State of Colorado et al. v. Google LLC (No. 20-cv-3715). I read the table of contents, the introduction (pp. 1-7), the section on including GenAI products (pp. 99-101), the section on the term of the judgment (pp. 219-222), the conclusion and the witness appendix. The findings of fact and the remedy-by-remedy analysis were not read in full.