A judge did not rule on AI Overviews. He ruled on the complaint.

FACT (INDEPENDENTLY VERIFIED). On 2026-09-30, U.S. District Judge Amit P. Mehta of the District of Columbia granted Google's motions to dismiss the amended complaints in two near-identical private antitrust suits brought by publishers against Google LLC and Alphabet, Inc.

• In Chegg, Inc. v. Google LLC (No. 25-cv-00543), Chegg alleged Google leveraged monopoly power in general search services to coerce publishers into supplying cost-free content, which Google then republished to its users.
• In Penske Media Corp. v. Google LLC (No. 25-cv-03192), Penske Media Corporation and nine affiliated publishers — Billboard Media, Deadline Hollywood, Fairchild Publishing, Gold Derby Media, The Hollywood Reporter, Indiewire Media, Rolling Stone, SheMedia and Variety Media — brought the same theory.
• Both pleadings alleged multiple Sherman Act violations and a common-law unjust-enrichment claim, and alleged that Google forced publishers to allow AI Overviews of their content or risk disappearing from search results, costing them referral traffic and revenue.

The judge heard the two motions together: he granted a consolidated oral-argument motion by Minute Order dated 2026-07-29 and heard argument on 2026-08-25 (104-page transcript), then took the matter under advisement.

INDEPENDENTLY VERIFIED via the opinion text. The Memorandum Opinion opens by characterising these as "two further entries in a series of private antitrust lawsuits brought against Defendant Google LLC arising from its dominance in the market for general search services."

FACT (INDEPENDENTLY VERIFIED). Every federal antitrust count and the state unjust-enrichment count were dismissed in both cases.

INDEPENDENTLY VERIFIED. Google moved to dismiss the amended complaints (Chegg ECF No. 19; Penske ECF No. 25). The PMC tying count was dismissed under Microsoft Corp. v. Bros. Computer Services Co., 253 F.3d at 85, 88, because the plaintiffs failed to plausibly allege that the tying and tied goods are two separate products and that consumers had no choice but to purchase the tied product.

The unjust-enrichment count fell not on its merits but on jurisdiction: "Having dismissed all federal claims in these actions, the court has discretion to either exercise supplemental jurisdiction over the remaining state law claims or decline to do so." The court declined, observing that in the typical case where all federal claims are dismissed the balance of factors points toward declining.

INDEPENDENTLY VERIFIED — an important precedent link. The court had already rejected substantively similar claims six months earlier in Helena World Chronicle, LLC v. Google LLC, No. 23-cv-03677 (APM), 2026 WL 787882 (D.D.C. Mar. 20, 2026), where publishers alleged a "monopoly broth" of anticompetitive acts. The court wrote: "The court has previously rejected similar claims of an agreement between Google and publishers, and Plaintiffs offer no persuasive reason to change course."

FACT (INDEPENDENTLY VERIFIED) — the legal mechanism. The court reasoned in four steps:

1. §1 requires an agreement. Section 1 "prohibits any 'contract, combination . . . or conspiracy, in restraint of trade or commerce'" and requires that "the challenged anticompetitive conduct stems from . . . an agreement, tacit or express." Plaintiffs "must plead facts showing concerted action between independent actors, not unilateral conduct."

2. What the plaintiffs actually pleaded was an expectation. "Plaintiffs have pleaded only that they have an 'expectation' that Google will send them search traffic if they make their content available for free." The operative passage, verbatim:

   "But an expectation is not an agreement. It is simply how a general search engine works." — Memorandum Opinion, Part IV.A.1 (citing United States v. Google LLC (Google Remedies), 803 F. Supp. 3d at 112–13)

   Immediate context: the court had just held that the plaintiffs "do not assert that Defendants offered terms they accepted; the parties negotiated quantity, duration, or exclusivity; or either side made any commitment," and that "merely incanting the phrase 'historical course of dealing' cannot plausibly convert two parties' independently self-interested conduct into a contract."

3. §2 cannot rescue it. Reciprocal dealing is "by definition, a two-sided practice" and "presupposes that a dealing relationship exists." "Absent any actual reciprocal dealing, there can be no exclusionary conduct."

4. The sympathy passage did not change the outcome. Verbatim:

   "The court does not treat Plaintiffs' alleged harms lightly. Nor is it unsympathetic to the situation publishers now find themselves in, and the knock-on consequences to journalists, educators, and other online creators whose content Google takes and repurposes without compensation. But the 'antitrust statutes' are not 'in any sense a substitute for [a] legislative body addressing questions of . . . economic dislocation caused by new innovation.' And '[t]o the extent that this case highlights any deficiencies . . . in the reach of the antitrust laws, that is an issue for Congress [or regulators] to consider.' The court 'is bound to apply the law only as it is written, not how the [c]ourt or any party believes it ought to be.'"

   Concluding line of that section: "Because the [P]laintiffs here have not nudged their [reciprocal dealing] claims across the line from conceivable to plausible, Counts I and II of their respective Amended Complaints must be dismissed."

FACT (INDEPENDENTLY VERIFIED). The earlier-referenced phrase "fail to get out of the starting gate" is the court's own words, applied to the reciprocal-dealing claims regardless of whether they receive per se or rule-of-reason treatment.

INTERPRETATION. The ruling is genuinely significant for AI product strategy, but its significance is narrower than the headline implies:

• The antitrust-theory risk is reduced, not removed. A court has now twice told publishers that "indexing plus AI summaries = coerced exchange" does not survive a motion to dismiss. That is a serious pleading bar for this framing.
• The risk is removed only at the pleading layer. Because the dismissal is without prejudice, a materially better-pleaded complaint alleging actual communications, commitments, or sudden pricing changes remains possible.
• It is emphatically not a fair-use ruling. Nothing here touches copyright. The separate fair-use litigation, including the Anthropic and New York Times cases in the SDNY, was not decided here and continues on its own track.
• It relocates the fight. The court expressly framed the deficiency in the antitrust laws as "an issue for Congress [or regulators] to consider" — an invitation to legislators, not a merits vindication of Google's product design.

FACT (COMPANY CLAIM / RESPONDING PARTIES). Neither Chegg nor Penske, nor Google, had publicly commented to Reuters at the time of reporting. Reuters reported Google has "denied any wrongdoing" in the litigation.

INDEPENDENTLY VERIFIED — market context. Reuters separately reported that in January 2026 Google told the court "PMC's preferred world, Google Search must be frozen in time" and that publishers can block indexing entirely, with "no referral traffic guarantees for indexed sites." The Verge reported that week that Google is paying around 100 publishers for contributions to AI Overviews, AI Mode and Gemini under a pilot programme.

FACT (INDEPENDENTLY VERIFIED): the pleadings as filed no longer support this theory. Specifically, the court's reasoning forecloses, absent new factual allegations:

• Relying on the mere fact that publishers allow crawling, to argue that crawling constitutes a bargain.
• Asserting Google's general encouragement of quality content as evidence of a commitment to send traffic.
• Using "historical course of dealing" as a substitute for pleaded terms of price or quantity.
• Treating a very broad "all online text content" market definition as plausible.

INTERPRETATION: Google and other answer-engine builders retain latitude to index and summarise publisher content without negotiating individual compensation contracts, at least against this antitrust theory.

1. Will Chegg or Penske replead, and if so, with what new alleged facts — actual communications, commitments, or price changes?
2. What does the Chegg Order's silence on prejudice mean operationally, and will the parties litigate it?
3. Will either side appeal, and to what effect on the without-prejudice posture?
4. Does the "issue for Congress" invitation get picked up in any pending legislative vehicle on AI content compensation?
5. Does the 2026 Google pilot paying roughly 100 publishers for AI Overviews contributions change the commercial picture, and would negotiated licensing terms paradoxically create the "agreements" this ruling found absent?
6. Do the standing and market-definition holdings migrate into publisher suits against other AI answer engines?

FACT (INDEPENDENTLY VERIFIED): both amended complaints are dismissed; the state unjust-enrichment claim is dismissed for want of supplemental jurisdiction; both Orders are final and appealable. No notice of appeal was found on either docket as of the last RECAP sync observed (2026-10-03).

PREDICTION (low confidence, explicitly speculative): the most likely near-term path is a decision by plaintiffs on whether to replead. Given that the same judge has now twice rejected the agreement theory, a straight repleading of the same facts is unlikely to succeed. A materially different complaint alleging actual communications, or a refiling on distinct theories, is more plausible.

PREDICTION (low confidence): the opinion's "issue for Congress" language is likely to be quoted in AI content-compensation debates, because it comes from a judge who found the plaintiffs' harm real but the available tool inadequate.

A judge who openly sympathised with publishers still dismissed their antitrust claims — and then told the plaintiffs, and the industry, exactly where to go next.

Three corrections belong in any coverage. First, the dismissal was without prejudice in the Penske case; the Chegg Order carries no prejudice designation. This was a Rule 12(b)(6)-style pleading ruling, not a merits win, and the suits can be repleaded. Second, this was a Sherman Act case. Judge Mehta decided nothing about copyright or fair use, and the separate fair-use litigation in the SDNY, including the Anthropic and New York Times matters, was not decided here. Third, the block that resonates most — "an expectation is not an agreement" — is legally narrow. It says the plaintiffs did not plead a bargain, not that AI Overviews are fair or lawful.

The durable signal is structural rather than celebratory: an AI system that acquires content automatically and universally, with no negotiated consideration, now has a judicial endorsement of the characterisation that this is how search works. The judge paired that holding with an explicit statement that the antitrust statutes "are not 'in any sense a substitute for [a] legislative body'" addressing the economic dislocation of new innovation. For publishers, the message is that the courtroom route on this theory has narrowed sharply. For the industry, it is that the remedy — if there is one — is being pointed at Capitol Hill.