On September 30, 2026, the U.S. District Court for the District of Columbia granted Google's motions and dismissed lawsuits brought by education company Chegg and Penske Media Corporation (PMC), whose properties include Rolling Stone and Variety. The publishers had argued that Google's "AI Overviews" and related features cut traffic to their sites and reduced their revenue.
The publishers claimed that Google's system forced them to let their content be used for AI features at no charge if they wanted to stay in Google Search. The court found that the complaints did not allege enough specific facts to establish an antitrust "dealing agreement" between Google and the publishers.
The arrangement described in the complaints is also not the same as the generative AI exclusion setting Google offers today. What the ruling shows is how hard it is to frame the relationship between Google and publishers that depend on search traffic as an unlawful deal under existing antitrust law.
Is the expectation of search traffic a "dealing agreement"?
Chegg sued Google and its parent company Alphabet on February 24, 2025, and PMC followed on September 12 of the same year. Judge Amit P. Mehta ruled jointly on Google's motions directed at the amended complaints. The 41-page opinion describes in detail how the two companies characterized the relationship between Google Search and publishers.
Publishers allow Google's crawler to access their articles and educational materials, and in return expect readers to arrive from search results. Turning those visits into advertising revenue and subscription fees has supported their content production.
But when Google began showing article excerpts and AI-generated answers directly on the results page, users could get the information they needed without visiting the original site. Chegg and PMC argued that this change unilaterally altered the existing arrangement: publishers provide content, and in exchange receive readers from search.
At the heart of the complaints was the concept of "reciprocal dealing," a situation in which parties are both buyers and sellers of each other, and one transaction is made conditional on another.
The companies characterized Google as providing search traffic on the condition that publishers supply content, free of charge, for uses such as generating AI answers and training models.
The court, however, found that the complaints lacked specific facts showing the underlying "agreement."
The complaints did not describe Google presenting terms that publishers accepted, any negotiation over quantities or duration, or any exchange showing that both sides had made a commitment.
The court did not say a written contract was necessarily required. It acknowledged that an implied agreement could arise from long-standing business practice, but concluded that the facts alleged did not reach that level.
Even if publishers expected to receive search traffic from Google and opened their sites to the crawler for that reason, that alone did not mean Google had promised to deliver a particular amount of traffic.
A publisher's heavy dependence on Google Search and the existence of a legal dealing agreement with Google are separate matters.
Four antitrust claims rejected, and state-law claims left undecided
The court dismissed the antitrust claims on different grounds, including the existence of an agreement, demand for separate products, antitrust standing, and market definition.
As for unjust enrichment, the court did not rule on the merits. Having dismissed the federal claims, it declined to continue hearing the state-law claims.
Summarizing Parts IV.A through IV.E of the September 30, 2026 opinion:
| Claim | Plaintiffs | Main reason given by the court |
|---|---|---|
| Reciprocal dealing between content and search traffic | Chegg, PMC | Insufficient specific facts showing an agreement. Part IV.A |
| Tying of Google Search and AI Overviews | PMC only | Failed to adequately show demand for separate products. Part IV.B |
| Maintenance of monopoly in general search | Chegg, PMC | Failed to adequately show antitrust standing in the general search market. Part IV.C |
| Monopolization and attempted monopolization in publishing markets | Chegg, PMC | Inadequate allegations on market definition and effects on competition in publishing and educational publishing. Part IV.D |
| Unjust enrichment from use of content | Chegg, PMC | After dismissing the federal claims, the court declined to exercise supplemental jurisdiction over the state-law claims. Part IV.E |
The source is the joint opinion in both cases, Part IV. The ruling examined whether each claim met the legal requirements based on the facts in the amended complaints. It is not a judgment that, after evidence was taken, determined whether or how much damage occurred.
In PMC's tying claim, the question was whether "Google Search" and "AI Overviews" are separate products.
PMC distinguished the two: traditional search sends users to other sites, while AI Overviews is meant to deliver answers within the search page.
The court, however, pointed to PMC's own allegation that users are satisfied with AI Overviews answers and do not click through to the original site. If what users want is the information itself, the difference between getting it from ordinary search results and from an AI summary did not sufficiently show independent demand for separate products.
The claim challenging the maintenance of a monopoly in general search had a different problem.
According to the court, the lost subscription revenue and inadequate compensation for use of content that the publishers cited were harms to the publishers' own businesses, not harms in the general search market itself.
The publishers did not sufficiently show a relationship, of the kind antitrust law protects, between Google's alleged restriction of competition in search and the harm they claimed to suffer.
The companies then pointed to effects on publishing markets, but this raised the question of how broadly the market should be defined.
PMC treated text published online as one broad market, but was found not to have adequately explained whether, for example, blogs or legal documents substitute for novels or news articles.
Chegg's proposed educational publishing market was likewise found to have a blurry boundary between information for learning and general information.
Google's strong position in general search and market power in a separate publishing market must be proven separately as a legal matter.
The unjust enrichment claims were handled differently.
The two companies had also argued that Google profited from using content without payment, citing unjust enrichment under California law.
But after dismissing all the federal claims, the court chose not to continue hearing these state-law claims.
The court therefore did not find that Google's use of the content was lawful under state law. Nor can any conclusion be drawn from this ruling about whether AI training or summarization is generally lawful under copyright law.
Google now lets sites opt out of generative AI features only
After the lawsuits were filed, Google began offering a setting that lets site owners remain in Search while excluding their sites from generative AI features.
According to the official Japanese help page, the "generative AI settings for Search" have been rolling out to websites worldwide since August 31, 2026. They cover AI Overviews, AI Mode, and generative AI features in Google Discover.
Site owners can use Search Console to choose whether their links and content are included in these generative AI features.
If a site opts out, content obtained by crawling it is no longer used to create the answers or displays of the covered generative AI features. Google says it does not use the setting itself as a signal for whether a site is listed or how it ranks in regular Google Search.
What Chegg and PMC objected to in their complaints was a situation in which blocking Google's crawling to refuse AI use meant also losing placement in regular search and the traffic that comes with it.
Under the current system, at least for the covered generative AI features, the two can be set separately.
It is therefore not appropriate to treat the circumstances at the time of filing, which the court described as background, as the current specification of Google Search.
Using content to generate AI answers is, however, a separate matter from training AI models themselves.
Google says the Search Console setting does not affect AI model training. To restrict use for training the models behind generative AI features in Search, it directs site owners to Google-Extended.
Excluding a site from generative AI features also means losing the chance for links to the site to appear within those features.
In exchange for keeping content out of AI answer generation, a site gives up opportunities to gain new readers from AI search. Even if regular search rankings are unaffected, there is no guarantee that traffic and revenue across the whole site will be maintained.
Nor is it appropriate to view this change in specifications as the reason the lawsuits were dismissed.
The main reasons the court gave were that the complaints' allegations on the existence of an agreement, market definition, antitrust standing and other points did not meet the necessary legal requirements.
The existence of a current opt-out mechanism is a separate question from whether publishers can seek compensation for past use of content or for harm already suffered.
The U.S. lawsuits and UK regulation start from different points
On June 3, 2026, the UK's Competition and Markets Authority (CMA) imposed a publisher-focused conduct requirement on Google's general search service.
The requirement calls on Google to, among other things, give publishers effective control over whether their content is used in Google Search's generative AI features. It also covers explaining how content is used and providing information that lets publishers understand how users engage with their content.
It also requires clear and accurate attribution of sources and access for users to the original content.
In the U.S. cases, Chegg and PMC applied the harm they claimed to existing antitrust requirements and sought relief from the court.
In the UK, by contrast, under a system built for digital markets, Google's general search service has been designated as having "strategic market status" (SMS), and specific obligations are imposed on particular conduct.
The CMA also explains that the SMS designation itself does not mean it has found Google to have engaged in anticompetitive conduct.
In other words, even where the U.S. suits dismissed here and the obligations imposed on Google in the UK address overlapping concerns, the legal mechanisms are fundamentally different.
The CMA's announcement also calls for options that let publishers keep their content from being used for further model training.
Google has been given a period to implement the changes, and the CMA requires reports on compliance, including data and metrics. The point at which obligations were imposed and the point at which all required measures are complete should be considered separately.
The UK approach emphasizes not only whether publishers can refuse AI use of their content, but also whether they can see what it is used for and how users engage with it.
With such information, publishers would find it easier to decide whether to allow AI use and what terms to seek.
However, measures imposed as a legal obligation in the UK and the Search Console setting Google has rolled out worldwide do not come from the same system.
Judge Mehta expressed understanding of the situation publishers face, while indicating that there are limits to resolving the economic changes brought by new technology through antitrust law alone.
The order can be appealed. To see how far publishers will be able to negotiate terms or compensation for AI use of their content, the effectiveness of opt-outs matters, but so does information that shows how much traffic and revenue AI search actually generates.