Why did Judge Mehta dismiss the Chegg and Penske lawsuits against Google?

In a consolidated opinion on September 30, 2026, US District Judge Amit Mehta granted Google's motion to dismiss in full, ruling that Chegg and Penske Media Corporation failed to plead a Sherman Act violation. Their theory was that Google coerced publishers into supplying content for AI Overviews for free under threat of disappearing from search. Mehta wrote that the plaintiffs had pleaded only an 'expectation' that Google would send them traffic in exchange for making content available, and that 'an expectation is not an agreement. It is simply how a general search engine works.' Without an agreement or an unlawful exclusionary act, there was no antitrust claim.

Did the court say AI Overviews don't harm publishers?

No. Mehta wrote that the court is not 'unsympathetic to the situation publishers now find themselves in' and acknowledged that AI search carries consequences for them. His point was jurisdictional: antitrust law cannot substitute for a legislature's decision about how to address the economic impact of a new innovation. The harm may be real; the Sherman Act is not the vehicle.

What options do publishers have after the dismissal?

Four main paths as of October 2026. Copyright: the Third Circuit's September 29 Thomson Reuters v. Ross ruling recognised an AI-training licensing market as a cognisable market and held that intermediate copying can infringe, which strengthens copyright claims over AI summaries. Licensing: The Information reported this week that Google is paying around 100 publishers in a pilot for contributions to AI Overviews, AI Mode and Gemini. Technical controls: Google-Extended and robots directives, Cloudflare's pay-per-crawl and Web Bot Auth, and bot management. Legislation: Mehta pointed explicitly to Congress; proposals on AI compensation exist in the US and the EU.

Can Chegg and Penske appeal?

Yes. The dismissal was with the motion granted in full; the plaintiffs can appeal to the D.C. Circuit. Chegg, which said AI Overviews collapsed its traffic and revenue, and Penske, parent of Rolling Stone, Variety and The Hollywood Reporter, had not announced next steps as of October 2, 2026. Because Mehta also authored the 2024 ruling that Google unlawfully monopolised search, his finding that this conduct is not exclusionary carries unusual weight on appeal.

Quick Answer

Chegg & Penske AI Overviews Suits Dismissed: What It Means

Published: October 2, 2026

The short answer

On September 30, 2026 Judge Amit Mehta — the judge who found Google an illegal search monopolist in 2024 — threw out the antitrust suits by Chegg and Penske Media over AI Overviews, in full. His reasoning fits in one sentence: publishers pleaded an expectation of traffic in exchange for free content, and “an expectation is not an agreement. It is simply how a general search engine works.” The court acknowledged publisher harm but said antitrust cannot substitute for legislation. Facts verified October 2, 2026.

The ruling

	Detail
Court	US District Court for the District of Columbia
Judge	Amit P. Mehta
Date	September 30, 2026 (reported October 1)
Plaintiffs	Chegg (filed February 2025); Penske Media Corporation — Rolling Stone, Variety, The Hollywood Reporter (filed 2025)
Claim	Google abused search monopoly power by coercing publishers to supply content for AI Overviews free, or lose search visibility; diverted traffic and revenue
Result	Motion to dismiss granted in full; consolidated opinion
Key line	”An expectation is not an agreement. It is simply how a general search engine works.”
On harm	Court “not unsympathetic,” but antitrust is not a substitute for a legislative response to “new innovation”
Significance	First US ruling on antitrust theories against AI search summaries

Why the theory failed

The publishers’ case needed an agreement or an exclusionary act. Their story was that there was an implicit bargain — crawl our pages, send us clicks — and that Google broke it by keeping the content and the user. Mehta’s answer is that no such bargain ever existed in law. Publishers make content available; a general search engine indexes it; traffic follows or does not. That is the product, not a contract. Repurposing indexed content into an AI summary may change the economics, but changing the economics of a market you lawfully dominate is not, by itself, exclusionary conduct under Section 2 of the Sherman Act.

The opinion matters more because of who wrote it. Mehta is not a Google-friendly judge: his August 2024 decision held Google unlawfully maintained a search monopoly, and he oversaw the remedies. His conclusion that this conduct falls outside antitrust is a strong signal that the D.C. Circuit will see it the same way.

What the court conceded

Mehta did not pretend the publishers were imagining things. He wrote that the court is not unsympathetic to the position publishers find themselves in — and the data is not in dispute. Chartbeat’s March 2026 analysis showed steep declines in search referrals to small sites; the Wall Street Journal has documented news-outlet traffic falling as AI answers absorb queries; andrew.ooo’s own Google Search Console shows the same post-I/O-2026 pattern. The court’s position is that this is a policy problem for Congress, not a Sherman Act problem for judges.

What publishers are left with

1. Copyright, not antitrust. The same week, the Third Circuit ruled in Thomson Reuters v. Ross that copying editorial content to train a competing AI product was not fair use, that intermediate copies can infringe even when they never surface in outputs, and that an emerging AI-training licensing market counts under the fourth fair-use factor. That is the doctrinal foundation publishers’ antitrust theory lacked. Copyright suits against AI summarisation now have an appellate decision to cite, even though Ross expressly declined to extend its reasoning to generative AI.

2. Licensing. The Information reported on September 29 that Google is paying around 100 publishers in a pilot for contributions to AI Overviews, AI Mode and Gemini. A licensing programme is exactly the “agreement” that Mehta said did not exist — and its existence is also evidence, under Ross, that a market exists. Expect publishers to argue both sides: no agreement for antitrust purposes, a market for copyright purposes.

3. Technical controls. Publishers can withhold content from AI training via Google-Extended, from AI Overviews grounding only by leaving search entirely — which is the coercion complaint in a different form. Beyond Google, Cloudflare’s Web Bot Auth and signed agents, pay-per-crawl and bot management give sites leverage over every other AI crawler, if not Googlebot.

4. Legislation. Mehta’s opinion is, in effect, a referral to Congress. Proposals for AI compensation to publishers exist in the US; the EU AI Act’s GPAI transparency and copyright obligations are already applicable. Nothing is imminent.

What it means for anyone who publishes on the web

Antitrust is closed as a route for AI-summary traffic loss, at least in the D.C. district and likely on appeal.
Optimise for citation, not just clicks. If AI answers are the surface, being the cited source inside them is the measurable outcome — AI-crawler and AI-user-agent hits in your server logs, not only Search Console clicks. See how to control AI shopping agents on your website for the agent side of that traffic.
Document provenance and licensing readiness. If the licensing market Ross recognised keeps growing, publishers with clean rights, structured data and measurable AI usage are the ones who get paid.

Last verified: October 2, 2026, against the dismissal opinion as reported by Reuters, The Verge and Ars Technica.