Google Wins AI Search Cases, but Publisher Questions Remain

Judge Amit Mehta has dismissed antitrust lawsuits by Chegg and Penske Media over Google’s AI Overviews, the search summaries the companies say divert readers from their sites. The same judge found in 2024 that Google unlawfully maintained a search monopoly. Those outcomes can coexist: the government’s earlier case examined conduct in specified markets, while the publishers’ later complaints had to connect Google’s alleged AI-search conduct to viable antitrust claims of their own.

Chegg and Penske alleged that Google used material from their sites to answer users’ questions within search results, leaving fewer people to click through and support their businesses. Chegg also argued that publishers faced a choice between making content available for AI use and losing visibility in search. Mehta rejected the amended complaints’ antitrust theories. The dismissals give Google a substantial win against these pleadings. They do not measure how many visits an AI Overview costs a publisher.

A publisher deciding whether to invest in search visibility still faces the business question behind the suits. A traffic chart can show a change in readership; it cannot, by itself, establish the prohibited conduct, market and injury needed for these antitrust claims. In court, the alleged exchange between publisher and search engine, the publishers’ position in the search market and their proposed markets for Google’s AI activity all came under scrutiny.

The Bargain the Publishers Described

Penske alleged that Google drew on publishers’ content and search dependence while AI-generated answers kept readers on Google’s results page. The publisher’s strongest practical argument is easy to understand. A site can make an article available for search discovery, then find that the search page gives a reader enough information to skip the visit. If that happens repeatedly, the publisher bears the cost of producing the article while losing opportunities to earn from the reader’s visit. How often AI Overviews produce that outcome for either plaintiff remains an open factual question.

The legal version of the argument was harder to sustain. The publishers described an implicit bargain: they supplied material for Google to index and expected search referrals in return. Mehta rejected the proposed agreement. “But an expectation is not an agreement,” he wrote, as quoted by Search Engine Journal in its account of the ruling. Treating ordinary indexing as a negotiated exchange would cast websites as parties to agreements they never made.

That reasoning keeps an expectation of referrals from doing the work of an agreement. It also exposes a problem the publishers’ theory struggled to address: a business can depend heavily on a system without having negotiated its terms. Economic dependence may explain why a change in search results hurts. On its own, it does not supply the agreement this claim required.

The apparent choice about AI use needs precision, too. Chegg alleged that publishers risked losing search visibility if they withheld content from AI use. Publishers can exclude pages from AI Overviews while retaining traditional search listings, Engadget reported. That option complicates a claim of outright exclusion from search. It may still leave a publisher weighing exposure on a results page against control over how its work appears there. The availability of a setting does not tell us whether the remaining listings deliver enough readers for that publisher.

Other theories encountered different obstacles. On Penske’s tying claim, the court found that the complaint had not plausibly established general search and AI Overviews as separate products with separate consumer demand. On claims involving unlawful maintenance of Google’s general-search monopoly, the publishers did not establish the antitrust standing required to challenge conduct in that market. Their proposed publishing markets also failed to support attempted-monopolization and monopoly-leveraging claims. A publisher can allege lost advertising revenue and still fall short of showing that it is entitled to bring a search-market antitrust claim.

Mehta acknowledged the alleged consequences for publishers and people who make online content. “The court does not treat Plaintiffs’ alleged harms lightly,” he wrote, as quoted by Ars Technica. That acknowledgment did not determine whether every claimed loss occurred. The dismissals addressed whether the amended complaints met the requirements of the claims they brought.

What the 2024 Search Case Decided

In August 2024, after the Justice Department’s case, Mehta found that Google had unlawfully maintained monopoly power in general search services and general search text advertising. Agreements affecting search defaults on devices and browsers were central to that case, as a competition-policy analysis of the liability decision explains. The decision at that stage addressed liability rather than remedies.

Chegg and Penske entered court with different allegations and a different task. They challenged what they said Google did with publishers’ content and referrals as AI answers appeared in search. The earlier finding makes their concern about Google’s power more intelligible: a site that relies on a dominant search engine for discovery has limited leverage over changes to its results page. The finding about agreements affecting search defaults does not establish an agreement to exchange publisher content for traffic. It also does not define the publishing markets in which the later complaints said Google was gaining power.

The comparison reaches its limit at the conduct each case examined. The government’s case tested how Google maintained its position in search, including agreements affecting where people search. The publishers’ cases asked whether Google’s use of content and the resulting alleged loss of referrals supported their own antitrust claims. A remedy aimed at search distribution would not, by itself, set a price for AI use of publisher content. That question would require its own legal or policy answer.

Traffic is a Separate Test

Chartbeat’s publisher network offers a measure of the pressure behind those claims. Google Search accounted for about 9% of pageviews across the network in July 2024 and 5% in July 2026, according to figures from Chartbeat’s 2026 Publisher Playbook described by Search Engine Journal. The network leans toward news and media sites, and the playbook does not specify a site count. Its year-over-year decline figures also leave unclear whether they measure pageviews or their share. The two-year figures here are shares of pageviews, not a count of visits lost by Chegg or Penske.

A smaller share of pageviews arriving through Google Search could reflect changes in search referrals, growth in other channels, or a combination of both. Even a measured drop in search visits would leave another question: how much came from AI Overviews rather than other changes in search or reader behavior? Chartbeat’s network figures do not isolate the effect of Overviews. Publishers have a reason to examine individual pages and referral patterns, but the aggregate trend cannot assign responsibility for either plaintiff’s alleged loss.

The UK Takes a Different Route

The UK’s Competition and Markets Authority has moved toward rules governing parts of search and AI without waiting for a publisher to win a US antitrust claim. After designating Google as holding a strategic position in search in October 2025, the CMA imposed publisher controls over AI features in June 2026. Its later proposal for Android and Chrome choice screens would include qualifying AI assistants alongside search services and require providers appearing on those screens to attribute publisher content clearly. The choice-screen changes remained proposals as of October 2; comments were due October 9.

The mechanisms address different points in the chain. A choice screen concerns which service a person starts with. Publisher controls concern how material is used. Attribution may help a reader find a source. None guarantees that a reader will click through after receiving an answer, or establishes how much traffic an AI Overview displaced. Google objected to repeating the choice prompt annually, arguing that it would annoy users and proposing a settings-based switch instead. That objection concerns how much friction the regulator should impose to make choice visible, rather than whether the publishers established the claims Mehta dismissed.

Penske’s dismissal was reported as without prejudice, a procedural opening that says little about the prospects of a revised complaint. Chegg’s precise procedural position should not be inferred from Penske’s order. For publishers considering what comes next, the court’s demand is concrete: identify an antitrust claim they are entitled to bring and plead the conduct and market that support it. For anyone assessing a policy response, the unfinished task is different: decide what control over content use and access to readers should look like even when that antitrust claim fails.