Skip to content

AI Search · The Spark

Publishers sued Google over AI Overviews and lost. The lever that actually moves citations was never in a courtroom.

On 30 September 2026 a US judge dismissed Chegg and Penske Media's antitrust suits against Google over AI Overviews. Here is what still earns AI citations.

On 30 September 2026, US District Judge Amit Mehta dismissed the antitrust lawsuits Chegg and Penske Media brought against Google over AI Overviews, ruling that a publisher's expectation of referral traffic is not an enforceable agreement. For anyone hoping a court would force Google to pay for or preserve search traffic, that lever just closed. It was never the lever that mattered.

TheWrap reported, in coverage published 1 October 2026, that Mehta granted Google's motions to dismiss in full, in a 41 page opinion. The claims, reciprocal dealing, tying, monopoly maintenance, and attempted monopolisation and leveraging, all failed on the pleadings. A further California unjust enrichment claim was dismissed on jurisdiction, with no ruling on its merits.

What the court actually said, and what it did not decide

Mehta found the publishers had not plausibly pleaded an antitrust violation, not that Google's use of their content is fair or lawful under other law.

On the central claim, Mehta wrote that the publishers had pleaded only an expectation that Google would send them traffic in exchange for making their content freely available, and that an expectation is not an agreement. WinBuzzer's coverage, published 5 October 2026, reports the same reasoning, and that Mehta separately rejected the argument that Google Search and AI Overviews are distinct products, the premise the tying claim depended on.

Per the same reporting, Mehta also said he is not unsympathetic to creators whose content Google takes and repurposes without compensation, and that antitrust law is not a substitute for legislation. Penske's dismissal was granted without prejudice, meaning that case can be refiled on narrower grounds. The legal door is not fully shut. It is a much narrower door than the original filings suggested.

Why this closes a shortcut, not the actual problem

The ruling settles a legal theory about traffic entitlement. It says nothing about whether your site gets crawled, structured correctly, or cited at all.

That distinction matters because a good deal of publisher strategy this year has quietly leaned on the hope that regulation or litigation would do the structural work instead. The UK took the opposite route in June 2026, when its Competition and Markets Authority ordered Google to let publishers opt out of AI Overviews and AI Mode without losing their ordinary search ranking. That is a binding remedy, not a dismissed complaint. It is also not the remedy most sites actually need.

An opt-out toggle controls whether your content can appear in an AI answer. It does nothing to make your content eligible to be cited in the first place if the crawler cannot read it, the schema does not validate, or the page has not changed since the last redesign. Our own August 2026 audit found six levers that actually move AI citation: crawlability, correct delivery, structure, schema, freshness and original data. None of them are legal questions, and none of them were waiting on a judge.

BETTING ON SOMEONE ELSE TOAntitrust claims dismissed on the pleadings,UK opt-out grants control, not citationEU review still gathering publisher feedbackWORKING THE SIX LEVERS YOUCrawlability and correct delivery, checkableSchema that validates, not just existsFreshness and original data, entirely yours
Two routes publishers bet on through 2026, and the one that was open the whole time

What we tell clients this week

Nothing about the dismissal changes the work ahead. It removes one more reason to wait for someone else to fix AI visibility for you.

In our experience, the sites that earned more citations after a redesign or a migration were never the ones with a complaint filed. They were the ones that ran the audit. We have written about what that audit actually finds: a content type header can make a page invisible to a crawler, a failure no court will ever rule on, and an llms.txt file does nothing for citation eligibility either, win or lose in court. Our schema markup playbook covers the lever Mehta's ruling has no bearing on at all: whether your structured data actually validates.

If you are a B2B site wondering whether this ruling changes anything for you, the practical answer is that it confirms what the six-lever audit already assumed. Run it, fix what it finds, and treat the schema and freshness work as the thing inside your control, because as of 30 September 2026, it is also the only thing anyone is going to fix on your behalf.

Mehta's ruling is a fact about a courtroom, not a fact about your site. The six levers were always the only ones with your name on them.

Frequently asked

Questions buyers ask about this

What did Judge Mehta rule in the Chegg and Penske Media cases against Google?

On 30 September 2026, US District Judge Amit Mehta dismissed both companies' antitrust claims over Google's AI Overviews, finding they had not adequately pleaded an antitrust violation. He wrote that a publisher's expectation of referral traffic is not an agreement.

Can Chegg or Penske Media refile their case against Google?

Penske's dismissal was granted without prejudice, so Penske can refile a narrower complaint. Reporting on Chegg's order is less consistent on this point, so treat that route as unconfirmed rather than assuming either case is fully closed.

Does the UK's AI Overviews opt-out order still apply after this US ruling?

Yes. The US dismissal is a separate antitrust case. The UK Competition and Markets Authority's June 2026 order, requiring Google to let publishers opt out of AI Overviews without losing search ranking, is a regulatory remedy and is unaffected by it.

If litigation and opt-outs do not guarantee AI citations, what does?

Our own August 2026 audit found six levers that actually move citation eligibility: crawlability, correct delivery, structure, schema, freshness and original data. None of them depend on a legal or regulatory outcome, which is why they are worth fixing regardless of how any court rules.

Working on a real engine? Start with a conversation.

Tell us where you are. We will tell you what we see and where we would start.