Google Loses Second Round to SerpApi: Court Sided Against It on 12 of 15 Points

A federal judge dismissed Google's amended DMCA complaint against SerpApi on 9 November 2026, the second dismissal in the scraping case. Our reading of the 19-page order finds the court sided with SerpApi on 12 of 15 contested points. Google has until 30 November to refile or the case ends with prejudice.

Vibhu Dhariwal
Cofounder & CTO, Digital Gratified
Published
Section
Industry news
Reading time
6 minutes
Card reading Industry news: 12 of 15 contested points the court decided for SerpApi

A federal judge in California has thrown out Google's amended complaint against SerpApi, which, according to Google's complaint, sells scraped Google results through a paid API. In an order dated 9 November 2026, Chief Judge Yvonne Gonzalez Rogers of the Northern District of California granted SerpApi's motion to dismiss for the second time. Google has 21 days, until 30 November, to file a new complaint, or the case will be dismissed with prejudice.

We read the 19-page order and logged every disputed point the court decided. Of 15 contested points, the court sided with SerpApi on 12 and with Google on 3. Google kept a narrow foothold on Reddit content, but every route it used to tie Google Search results to copyrighted material failed.

What the court decided

  • Both claims dismissed again. Google's case rests on two anti-circumvention provisions of the DMCA, 17 U.S.C. § 1201(a)(1)(A) and § 1201(a)(2). It brought no breach of contract claim, although the order notes Google's Terms of Service prohibit automated access.
  • The missing link is the search result itself. For the Reddit content, the "major content provider" deal and the Privacy Policy, the court found no allegation that the results SerpApi scraped actually contained that content.
  • Leave to amend is narrow. Google may refile only to fix the identified gaps, and the order says it must do so without contradicting its earlier allegations. Google must also send the court a redline showing what changed.
  • Discovery stays frozen. It has been on hold since the first dismissal on 20 July 2026 and remains stayed until any new complaint and motion to dismiss are resolved.
  • Scraping was not declared lawful. This is a ruling on whether Google pleaded enough facts. The court did not decide whether SerpApi's methods are legal.

The scorecard: 15 points, 3 for Google

To state a claim, Google had to show that SearchGuard, the JavaScript challenge it launched in January 2025 to block automated queries, controls access to copyrighted work, and that the copyright owner authorised it. Google's amended complaint offered four sources of that authorisation plus a catch-all for other content. The table maps each one against the tests the court applied.

Content Google relied onPleaded as copyrightable?Shown in scraped results?Authorisation pleaded?Authoriser had the rights?
Reddit licenceNoNoYes (Reddit's request to Google)No, for users' content
"Prominent licensing partner"Not decidedNot decidedNoYes (holder treated as owner)
"Major content provider"NoNoNoNo
Google Privacy PolicyNoNoNoNot decided
Lyrics, business images, user contentNot decidedNot decidedNoNot decided
Our coding of the 9 November 2026 order (Dkt. 62, N.D. Cal. case 4:25-cv-10826-YGR). Google's third win, not shown in the grid: the court agreed it need not attach or quote its licence agreements verbatim. Source: court order.

The three points Google won matter for what comes next. The court accepted that Reddit's request to Google to use technical measures against scrapers could count as authorisation, even though it was not a written contract term. It treated the unnamed partner who "holds" copyrights as an owner. And it agreed Google does not have to attach the contracts themselves, as long as it describes their terms in enough detail.

What sank the complaint was detail. The two licensing agreements were described without naming the parties, the content or the relevant terms. The Privacy Policy, which the court treated as part of the complaint, contains no terms about licensed content at all. On Reddit, Google never said what the licensed content was or showed that it appeared in the results SerpApi pulled.

How the case got here

Annotated timeline of Google v. SerpApi from the December 2025 filing to the 30 November 2026 refile deadline, with both dismissals marked
Dates from the court order and Google's announcement of the suit; day counts are ours. Source: N.D. Cal. order, 9 November 2026, and Google.

Google filed on 19 December 2025. Its general counsel, Halimah DeLaine Prado, wrote that day that SerpApi resells content Google licenses, such as Knowledge Panel images and real-time data, and that the activity "has increased dramatically over the past year." According to the complaint as the court summarises it, SearchGuard blocked SerpApi when it launched in January 2025, but SerpApi then found ways around it, including solving a challenge once and sharing the resulting authorisation across other automated browsers.

The first dismissal on 20 July 2026 shut one door permanently: Google cannot base claims on results that contain no copyrighted content. It left open results that do contain such content, which is why the amended complaint of 10 August leaned on licensing deals. The new order says those deals were not described well enough to connect them to the scraped pages.

The order also draws a contrast with Reddit v. SerpApi, a 2026 case in federal court in New York, where Reddit's DMCA claims survived a motion to dismiss because Reddit described the scraped snippets as creative contributions from users and explained how its user agreement let it protect them. Google's pleading lacked those specifics. In its own post on 10 November, SerpApi called this the second dismissal of Google's claims and said it is ready to keep defending the case. We found no public Google comment on the new order.

What this means for SEO teams

Many rank trackers, AI visibility tools and SERP datasets depend on automated collection from Google, so this case is a test of how far Google can go in court. For now, the answer is: not as far as it hoped. That does not make collection easier in practice. Google's technical measures continue regardless of the litigation, and we have already seen rank trackers served altered results pages and organic links hidden behind google.com/goto redirects.

  • Mark 30 November. If Google refiles, expect named licensing partners and specific examples of scraped results carrying licensed content. If it does not, the case ends with prejudice.
  • Do not read this as a green light. The order rules only on pleading. Reddit's own DMCA claims against the same company survived a motion to dismiss in New York, and Google's Terms of Service still bar automated access.
  • Ask your data vendors how they collect. A tool that depends on getting around SearchGuard could lose coverage overnight if a court or a technical change goes against it. Ask what the fallback is.
  • Anchor reporting in first-party data. Use Search Console clicks and impressions as the baseline and treat third-party rankings as directional, especially for sudden one-day swings.
  • Watch Reddit content. Reddit's licensing relationship with Google is now part of a court record about who controls access to that content in search, alongside Reddit's own test of logged-in only access.

Methodology

On 16 November 2026 we downloaded the order (Docket 62, filed 9 November 2026) from the public court record and read all 19 pages. We counted a point as contested when the order addressed an argument from SerpApi and stated whether it succeeded. We logged 15 such points across the five content categories and the procedural question of attaching contracts. Points the court expressly declined to consider, such as arguments SerpApi tried to incorporate from earlier briefs, were excluded. Day counts run from the filing date of 19 December 2025, taken from the order.

Conclusion

Google has lost the same argument twice: it has not shown, in its pleadings, that the search results SerpApi scrapes carry copyrighted work that a rights holder authorised Google to lock behind SearchGuard. The court left a path open, but it runs through naming licence partners and pointing to real scraped pages, which Google has so far avoided doing. The next date to watch is 30 November.

Questions and answers

Did the court rule that scraping Google is legal?

No. The 9 November 2026 order decides only whether Google pleaded enough facts for its two DMCA anti-circumvention claims. It does not decide whether SerpApi's collection methods are lawful.

What is SearchGuard?

According to Google's amended complaint as summarised in the order, SearchGuard is a JavaScript challenge Google launched in January 2025 that asks browsers sending queries from unrecognised sources to prove they are real users.

What happens if Google does not refile by 30 November?

The order says a failure to file an amended complaint within 21 days of 9 November 2026 will result in dismissal of the action with prejudice.

Why did the Reddit licence not save Google's claims?

The court accepted that Reddit's request to Google could count as authorisation, but found Google did not describe the Reddit content as creative, did not show it appeared in the results SerpApi scraped, and did not show Reddit could authorise protection of its users' content.

More releases

  1. Updates

    Gemini 4 Carbon Reaches AI Pro Users in AI Mode, but Stays Out of All 30 EEA Countries

    Google's Rajan Patel says Gemini 4 Carbon appearing in AI Mode for AI Pro subscribers is not a bug, though Google's help page still lists it for Ultra only. Our check of Google's country lists finds Carbon in AI Mode missing from all 30 EEA countries, despite both paid plans being sold there.

  2. Updates

    Search Console API Adds Branded, Multimodal and Multi-Country Filters, 361 Days After UI

    Google's Search Analytics API can now return branded and non-branded data, multimodal search performance from Lens and Circle to Search, and multi-country or multi-device filters in one request. The branded filter reached the API 361 days after its announcement, and Google's API specification caps each filter list at 1,000 values.

  3. Studies

    Does Google Pause Updates for the Holidays? One Ran on 51% of Peak-Season Days

    Google says it tries to avoid ranking updates from late November to mid-December. Its own Search Status Dashboard shows a notable update rolling out on 53 of 104 days in that window from 2022 to 2025, against 30% of days across the year. The one gap since 2022: no launches in late November.

Every release

Tell us the site and the goals.

Rankings in Google and visibility in AI answers, from one monthly plan. One call is enough to say which one fits. Email works just as well.