Antitrust Remedies in United States v. Google: AI and the Evolving Search Market

Following the August 5, 2024, decision in United States v. Google where the D.C. District Court found Google to be a monopolist,^[1]Judge Amit Mehta issued what would end up being the first remedies opinion on September 2, 2025.^[2]The parties vigorously debated the meaning of that opinion and submitted competing final proposed judgments, which led Judge Mehta to issue a final remedies decision and judgment on December 5, 2025.^[3]

The decision establishes a six-year framework designed to promote competition in the general search engine market. The court’s approach seeks to open up distribution and scale advantages built by what it found was Google’s unlawful exclusivity while avoiding product redesign mandates, forced breakups, or disclosures that might compromise Google’s intellectual property.

GENAI: A NEW DIMENSION

When the case began in 2020, generative artificial intelligence (GenAI) was virtually absent from the market and was left out of the debate over defining the relevant market. However, by the time the case entered the remedies phase after the 2024 liability decision, the rapid rise and widespread use of GenAI, including for various search functions, became a key consideration. GenAI tools that answer broad, information‑seeking prompts had become a plausible substitute for many search tasks. 

Google agreed to include GenAI tools within the scope of the remedies, but despite Google’s stated willingness, its proposed remedies regarding AI were far short of what the U.S. and state plaintiffs envisioned. Google argued that GenAI was not part of the court’s initial liability finding, so any remedies regulating licensing of Google GenAI products should be narrower than remedies regulating Search or Chrome.

The judge generally rejected Google’s approach and instead adopted the plaintiffs’ more robust remedies. The judgment which was entered covers Google’s GenAI tool (Gemini), any successors, and any other Google products that fit the GenAI definition. The court treats GenAI products like Search wherever they function in a similar fashion so that Google cannot use its monopoly power in search to box out GenAI challengers.

In short, the court noted that the remedies are designed to stop Google from simply reusing its old tactics in the dynamic AI market.

THE NEW RULES

In addition to the treatment of GenAI, two conduct refinements are most noteworthy. First, the court adopts a “Google shall not condition” rule which prevents Google from making access, payments, or favorable terms for one product contingent on using another Google product, setting it as the default, or sidelining a competitor.

Second, any agreement setting Google as the default search engine must end after one year without any automatic renewal provisions. This ensures genuine annual rebidding and prevents de facto multi‑year lock‑ups.

The rules apply at a device‑by‑device and access‑point‑by‑access‑point level, which means partners don’t have to make one all‑or‑nothing choice about using Google across their entire product line. Device makers, carriers, browser developers, and Apple can choose different default preferences for each device and each access point, such as browser modes, voice assistants, and GenAI apps.

Data, Not Secrets: Targeted Disclosures

To curb advantages gained from unlawful exclusivity, the final judgment requires Google to make various data disclosures to “Qualified Competitors”—companies approved to receive the benefits of the final judgment because they can demonstrate that they will use such benefits to compete with Google in search or AI. Under the judgment, approved Qualified Competitors can access limited web search index metadata and defined “user‑side” datasets.

At the same time, the court did not require Google to hand over its intellectual property, including specific details about algorithms, ranking signals, and trained models.

Syndication as a Launchpad

A five-year syndication of Google search results and search text ads pathway in the final judgment provides a temporary, structured means for competitors to serve users and earn revenue while they build their own search and AI systems.

While competitors cannot use the syndicated content to train large language models, the syndication option enables them to temporarily and quickly use Google’s search results to produce interactive results pages while training their own GenAI models with other data sources.

Who’s Watching: Oversight And Compliance

Oversight is strengthened through the Technical Committee, which can recommend data-security standards, advise on reducing competitors’ reliance on Google’s syndication over time, and help standardize license templates for data sharing and syndication.

It will be interesting to watch the remedies develop over the coming months and years as AI matures and the effectiveness—or lack of effectiveness—of the court’s remedies becomes manifest.

Law clerk Atiya Dosani also contributed to this blog post. 

---

[1] United States v. Google LLC, 747 F. Supp. 3d 1, 187 (D.D.C. 2024).

[2] United States v. Google LLC, No. 20-CV-3010 (APM), 2025 WL 2523010 (D.D.C. Sept. 2, 2025).

[3] United States v. Google LLC, No. 20-CV-3010 (APM), 2025 WL 3496448 (D.D.C. Dec. 5, 2025).