Is scraping Google's search results a DMCA violation? A judge said not by itself — so Google's amended complaint pivots to licensed content

Where the case stands (docket read 2026-09-30)

We re-read the CourtListener docket on 2026-09-30. Every entry after the amended complaint (entry 45, 2026-08-10) is below. Entries 46, 48, 49, 55, 56 and 57 are free PDFs on CourtListener and we read them; entries 47, 51, 52, 54, 58 and 59 exist there only as docket text, and we quote that text rather than the documents. The two briefs on the motion to dismiss after SerpApi's opening brief (Google's opposition, entry 58; SerpApi's reply, entry 59) are among the ones we have not read.

SerpApi's motion to dismiss opens: "This case is a shadow of its former self. Google began by seeking to block access to all its public search results. The Court rightly rejected Google's attempt to convert the DMCA into a general-purpose anti-scraping statute, dismissing with prejudice any claim for search results not involving copyrighted works." Its argument is that the amended complaint names four sources of authority for SearchGuard (an agreement with an unnamed licensing partner, one with an unnamed content partner, the Reddit agreement, and Google's own privacy policy) and quotes the operative terms of none of them: "Google chose not to place before the Court the three license agreements it proffers as the cure for its prior pleading deficiencies. Nor does it quote the operative terms." SerpApi's brief quotes the July order back at Google: without the text, "the Court cannot reasonably infer that such license agreements grant Google authorization to deploy technological measures to control access to the copyrighted content at issue."

SerpApi's separate motion asked the court to order Google to serve those three agreements so their text could be read against the motion to dismiss. The court denied it on 2026-09-15 without holding the hearing, in a three-page order. The two reasons, in the order's words: "First, the fact that SerpApi has already responded to the AC by filing a motion to dismiss, see Dkt. No. 49, invalidates its contention that it cannot adequately respond to the AC without the licensing agreements at issue. Second, SerpApi has not cited a single case in which a court has ordered the production of documents mentioned in a complaint on the ground that they are incorporated by reference into the complaint. Nor has SerpApi cited any authority that the incorporation-by-reference doctrine can be used as a discovery mechanism." The order adds that the doctrine "can be used to allow the Court to evaluate known documents referenced in a complaint," but that "no legal authority exists for the proposition that normal procedural, including discovery, processes should be circumvented." It also restates the July holding the amended complaint has to overcome: allegations that Google licenses the content "were insufficient, without more, to raise the inference that copyright owners authorized it to deploy technological measures to control access to their copyrighted content," and notes that discovery is stayed until the motion to dismiss is resolved. So the motion to dismiss goes to the 2026-10-13 hearing with Google's description of the agreements on the record and the agreements themselves not.

Corrections made on September 30. The 2026-09-22 version of this page said the entry 57 order was on PACER only and that we had not read it; the PDF is free on CourtListener and is quoted above. It said the docket text for entry 58 did not name the filer; the docket text reads "filed by Google LLC." Entries 47, 48 and 50 to 54 were left out of the table; they are in it now.

What happened, from the docket

Google sued SerpApi — the API vendor that sells structured, scraped Google results at scale — in the Northern District of California on 2025-12-19, pleading two claims under the DMCA's anti-circumvention provisions (17 U.S.C. § 1201(a)(1)(A) and § 1201(a)(2)): that SerpApi circumvents SearchGuard, Google's anti-bot access control, and traffics in the means to do so. On 2026-07-20, the court granted SerpApi's motion to dismiss — partly with leave to amend, and partly without: where the scraped results pages contain no copyrighted content, the DMCA claims "fail as a matter of law" and "amendment would be futile." Google's amended complaint, filed on the court's 2026-08-10 deadline, re-pleads both § 1201 claims — now built on the copyrighted material Google licenses into its results.

The licensed-content theory, in Google's own words

The amended complaint's clearest new material is contractual. Google describes its licensing agreement with Reddit as one that "directs Google not to enable third parties to extract and independently commercialize the licensed content," and alleges its licensing partners have "expressly requested that Google prevent unauthorized access to licensed content." It also leans on Reddit's separate, ongoing suit against SerpApi (Reddit Inc. v. SerpApi LLC, No. 25-cv-8736, S.D.N.Y.), quoting Reddit's own filing that it "has always authorized its partners, including Google, to protect [the licensed] content from misuse." Knowledge-panel photographs, music lyrics, and licensed user-generated content are the copyrighted components Google now says SearchGuard protects.

Why this docket matters for AI-visibility measurement

Most tools that report on Google AI Overviews and AI Mode visibility do not get that data from Google — the official Search Console surface for generative-AI performance has no API, no query data, and no click data. The industry's working substitute is scraped SERPs, and SerpApi is one of the largest suppliers of exactly that. This case is drawing the legal boundary around the supply chain.

• The July ruling is, so far, a structural point in scrapers' favor: circumventing an anti-bot wall is not by itself a DMCA violation when what's behind the wall isn't copyrighted. That portion is dead without leave to amend.
• Google's counter-move makes content licensing the battleground: if results are laced with licensed copyrighted material, the same scrape may look different legally. AI-visibility vendors whose dashboards depend on scraped AI Overviews inherit that uncertainty either way the amended complaint resolves.
• The Reddit angle cuts across engines: content owners who license to one AI surface are asserting the right to keep that content out of everyone else's pipelines — the same tension behind lost citations when licensing deals change. Our audits sit outside this fight — they query each engine's own API and read GSC through the official API — but any business comparing AI-visibility vendors should ask where the Google numbers come from. Check what the engines themselves say about your business rather than relying only on scraped dashboards.

No merits ruling exists on the amended complaint. Briefing on the motion to dismiss closed with SerpApi's reply on 2026-09-29 (entry 59); the hearing is set for 2026-10-13 at 2:00 PM in Oakland. The next edit to this page follows the order on that motion, read from the docket, not from either side's press coverage.