SerpApi filed a 23-page motion on August 24, 2026 asking a federal judge to dismiss Google's rewritten copyright complaint with prejudice, arguing the search company has built its entire remaining case on licence agreements it has never placed before the court.

The filing, docketed as Document 49 in Case No. 4:25-cv-10826-YGR in the United States District Court for the Northern District of California, is the second time the Austin, Texas company has asked Chief Judge Yvonne Gonzalez Rogers to throw out claims brought by Google under the Digital Millennium Copyright Act. The first attempt worked. A hearing on the new motion is set for September 29, 2026 at 2:00 p.m. in Courtroom 1 on the fourth floor.

What remains of the dispute is a fraction of what Google filed in December. The opening line of the motion puts it bluntly: the case is now, in SerpApi's words, "a shadow of its former self."

What the July order left standing

Judge Gonzalez Rogers dismissed every claim in Google's original complaint on July 20, 2026, splitting the ruling into two parts. Claims covering search results that do not involve copyrighted works were dismissed with prejudice, meaning they cannot return. Claims relating to licensed copyrighted content were dismissed without prejudice, on the ground that Google had not alleged that its licence agreements gave it copyright owner authorisation to deploy SearchGuard, the branding Google uses for its own anti-bot system.

That left Google with two tasks on amendment, as the motion frames them: identify the copyrighted works at issue, and plausibly allege that the owners of those works authorised the access control. Google filed its amended complaint on August 10, 2026, the final day of the 21-day window. SerpApi's position is that neither task was completed.

The amended complaint no longer reaches the links that make up the bulk of a results page, nor the AI-generated Gemini answers Google displays to users, nor copyrightable material Google itself scrapes from third-party websites. Those categories sit outside the pleading now. What is left is unspecified content that Google licenses on a non-exclusive basis for display alongside results.

How much Google cut

A Litera redline comparing the December 19, 2025 complaint with the August 10, 2026 version, generated at 5:33 p.m. on the day of filing and provided by SerpApi, counts 163 total changes: 93 additions, 65 deletions, two moved passages and one newly embedded graphic.

The deletions are the more consequential half. Google's original pleading named Google Shopping, where merchants supply product pictures, and Google Maps, where the company displays terrestrial imagery, business-supplied photographs and user-generated reviews, as examples of search surfaces carrying licensed copyrighted material. Both product theories are struck from the amended text. In their place sits a single revised sentence covering music lyrics, merchant-supplied product and business images and descriptions, and user-generated content generally.

The paragraph on SerpApi's operations was renumbered but survives largely intact, including Google's estimate that the company sends hundreds of millions of artificial search requests each day, a volume the complaint says has grown by as much as 25,000 percent over two years. The new embedded graphic is a Knowledge Panel screenshot for a query on Babe Ruth, added alongside the Willie Mays panel that appeared in the December version.

Google's counsel of record remains David H. Kramer of Wilson Sonsini Goodrich and Rosati in Palo Alto. SerpApi's motion is signed by Colin R. Kass of Proskauer Rose in Washington, with David A. Munkittrick and Joseph E. Clark of the firm's New York office and Heather Tovar of Ad Astra Law Group in San Francisco also listed. The February 20, 2026 motion that produced the July dismissal was led by Clark; Kass signs this one.

The four claimed sources of authorisation

Google's amended complaint points to four places where, it says, copyright owner permission to run SearchGuard can be found. The motion attacks each in turn, and its statement of issues is organised around them: an agreement with an unidentified licensing partner, an agreement with an unidentified content provider, the Reddit agreement, and Google's own Privacy Policy.

The threshold objection applies to the first three at once. Google has not filed any of those agreements with the court, and does not quote their operative terms. It offers its own characterisation of what the contracts require, then asks the court to infer further implied rights beyond that characterisation. SerpApi's lawyers call these "legal conclusions masquerading as facts," invoking the Supreme Court's 2007 decision in Bell Atlantic Corp. v. Twombly, which held that courts need not accept a legal conclusion dressed as a factual allegation.

The July order made the same point in different words. Without the contractual text, the court wrote, it cannot reasonably infer that the licence agreements grant Google authorisation to deploy technological measures over the copyrighted content at issue. The motion argues the amended complaint reproduces that gap rather than closing it.

The unidentified licensing partner

Google alleges that a prominent licensing partner, holding copyrights to millions of works under agreements dating back to 2017, obligated it to use commercially reasonable efforts to safeguard licensed content against unauthorised third-party access.

Two problems, according to the filing. First, holding a copyright and owning one are different things. A distributor or aggregator can hold millions of copyrights for purposes of sublicensing and sale without being the statutory owner of any of them, and Section 1201 requires authority from the owner. The motion treats Google's choice of the word "holds" as deliberate rather than careless.

Second, nothing establishes that automated traffic to a public website counts as unauthorised access under that agreement. Reaching google.com requires no account, no password and no credential. In ordinary usage, the motion argues, unauthorised access describes hacking of internal systems, closer to the meaning the Computer Fraud and Abuse Act ascribes to the phrase. A licensor might reasonably want assurance that servers housing its catalogue are protected from intrusion without saying anything at all about bot detection on a public results page.

The filing then reduces the question to four possibilities, each of which it says defeats the claim. Either the agreement is silent on automated access, or it does not reach automated access, or it leaves the copyright owner to define unauthorised access, or it delegates that decision to Google. Silence and non-coverage confer nothing. If the owner defines the term, Google would have needed to allege that the owner exercised that authority, which it did not. Delegated authority was not alleged either.

Underneath sits a structural argument. Google prohibits automated access in its own Terms of Service, a point the amended complaint makes at paragraph 27. The motion contends Google then carries that definition forward as though every agreement it holds with anyone incorporates it, calling the move "sleight-of-hand, not well-pled fact." A unilateral decision by Google to block bots on its own site is not, on this reading, permission granted by a copyright owner.

The unidentified content provider

This source occupies a single sentence in the amended complaint. Google says a major content provider requires that licensed content not be available for download by third parties, and treats that requirement as authorising technical access controls.

The motion's response turns on a distinction that runs through DMCA case law: "An anti-download provision is not an anti-access provision." Downloading is copying, and copying falls under Section 106 of the Copyright Act, the ordinary infringement provision. Section 1201(a) governs access. The filing cites Hattler v. Ashton, a 2017 decision from the Central District of California reading the statute to exclude technologies that permit access while restricting copying, and the Ninth Circuit's 2010 ruling in MDY Industries v. Blizzard, which distinguished measures targeting access from measures regulating copying.

There is also a plausibility argument. Read literally, an obligation preventing all third-party downloads would put Google in breach whenever a user saves an image or takes a screenshot of a results page. The motion notes that Google's own counsel copied the Willie Mays and Babe Ruth pictures reproduced in the pleading. More sensible readings exist, the filing suggests: a bar on Google sublicensing or reselling the content, or a bar on placing a download button next to it.

The Reddit agreement

The Reddit portion is the most heavily worked section of the motion, running across three subheadings.

Google does not contend that Reddit owns the copyright in its users' posts. It asserts instead that Reddit has the power to authorise access controls both for its own works and for works its users licensed to it. The motion treats that as another legal conclusion with no cited agreement behind it, and points out that Google is not in contractual privity with Reddit's users at all.

The filing quotes the Reddit User Agreement, last revised May 26, 2026, which leaves ownership with users and grants Reddit a worldwide, royalty-free, perpetual, irrevocable, non-exclusive, transferable and sublicensable licence to use, copy, modify, distribute and display their content. Those are exploitation rights. None of them, the motion argues, addresses who may impose a technological access condition or lets Reddit delegate that power downstream.

Reddit's Public Content Policy, updated May 29, 2025, cuts the other way still further. It states that most of the platform is public and accessible to everyone, without an account, and that this is intentional. The motion's characterisation is direct: "This is the opposite of a public access restriction."

A separate argument concerns what actually appears on google.com. The amended complaint does not identify a single Reddit post appearing in any search result, or any portion of a result scraped by SerpApi. Facts and information are not copyrightable, and short excerpts may lack the originality copyright requires. The filing leans on the parallel New York decision, Reddit, Inc. v. SerpApi LLC, in which the court observed that the Reddit-derived material on Google's pages consists of snippets that do not appear to exceed a few sentences, and framed the relevant question as whether those snippets qualify for protection rather than whether the full posts do.

Then there is the chain of characterisation. Google does not paraphrase its Reddit agreement directly. It quotes Reddit's separate amended complaint against SerpApi, in which Reddit characterises the agreement, and alters that characterisation by bracketing the word "licensed." Neither company has put the contract before any court. The motion declines to accept a characterisation of a characterisation of an unquoted agreement.

Finally, Google alleges that when Reddit suspected scrapers were taking licensed content, it specifically asked Google to deploy technical measures. The motion treats that as parol evidence, inadmissible to prove the rights conveyed under a written agreement unless the court first finds the agreement ambiguous.

The Privacy Policy

Google's fourth source is its Privacy Policy, which the amended complaint says commits the company to protecting user-uploaded and licensed content from unauthorised access.

The motion attaches the Policy as an exhibit and argues it does the opposite of what Google needs. It is not an agreement, requires no consent, does not mention copyright or SearchGuard, and governs user information: names, passwords, phone numbers, payment details, email addresses, device data, location and the content users create or receive inside Google services such as Gmail and YouTube. The only unauthorised access the Policy addresses concerns information Google holds about its users, described through encryption, Security Checkup, two-step verification and physical controls on Google's systems. The filing calls this ordinary data-security language.

A footnote lands a narrower point. Google's amended complaint says the Policy covers licensed content users supply. The motion observes that the only appearance of the word "license" in the Policy is "driver's license."

The copyright licence users grant, the filing argues, sits in Google's Terms of Service, which conveys a non-exclusive right to display content and contains no authorisation or obligation to run anti-bot technology. The Privacy Policy does not fill that void.

The motion also turns Google's own litigation history against it, citing a 2023 filing in J.L. v. Google LLC in the same district, made by the same counsel, in which Google argued that web scraping does not collect personal information because publicly available information falls outside the definition.

A discrepancy in the count

The materials SerpApi circulated contain a mismatch worth noting. General Counsel Chad Anson, in an on-the-record statement, describes Google's amended complaint as resting on four agreements that have neither been shown to the court nor quoted in the pleading. The motion itself identifies four sources of claimed authorisation but only three agreements. The fourth, the Privacy Policy, is a published document, and SerpApi filed it as an exhibit to the Munkittrick declaration alongside Google's Terms of Service.

The distinction matters for how the argument is read. Three contracts are absent from the record. The fourth source is present, and the objection to it is not that the court cannot see it but that its text says nothing about copyright.

Anson's statement restates the company's position without change since December: SerpApi provides "programmatic access to search results that are publicly visible" to anyone with a standard browser. Copyright law, he added, "was not written to let a platform control who reads a public page." Anson published a similar framing on January 23, 2026, weeks after joining the company.

Arguments preserved for appeal

Section III of the motion renews five grounds the court rejected in July, stated compactly for the record rather than reargued.

Google, as a non-exclusive licensee and website operator rather than a copyright owner, falls outside the DMCA's zone of interests. SearchGuard is not an effective technological measure controlling access to a copyrighted work, for three reasons: the underlying content stays reachable elsewhere on the open internet, the system is designed to stop automated access regardless of whether copyrighted material is present, and no measure is required to reach the work, a point drawn from the Ninth Circuit's 2022 decision in hiQ Labs, Inc. v. LinkedIn Corp. Google has not alleged circumvention as Congress defined it. The trafficking claim fails because most of what SerpApi retrieves is not alleged to be copyrighted and because licensed Knowledge Panel content is a small fraction of what its customers receive. And Section 1203(a) requires actual injury caused by a Section 1201 violation, which the motion says Google has not pleaded.

None of these was cured by amendment, the filing states.

The Babe Ruth problem

A footnote raises a question about the only specific works Google identifies. The amended complaint's sole alleged instances of protected content are the licensed images of Babe Ruth and Willie Mays displayed in Knowledge Panels.

Google does not allege that either image is copyrighted. The Babe Ruth photograph dates to 1921 and is in the public domain. The Willie Mays photograph dates to 1951, a period when maintaining copyright required affirmative renewal steps, none of which the complaint alleges were taken. Nor does Google allege that either image falls under any of its four claimed sources of authority.

For a pleading that had to identify copyrighted works to survive, the motion treats that as a gap at the centre rather than at the margin.

Two courts, two directions

The California and New York cases are now moving apart. In late July 2026, Judge Paul Engelmayer allowed Reddit to proceed with claims that SerpApi and Perplexity wrongly obtained copyrighted posts from Google results pages, while Judge Gonzalez Rogers dismissed Google's complaint in full days earlier.

The motion argues the two cases sit in different procedural postures. Reddit pleaded allegations about the copyrighted nature of specific posts, and that court deferred the protectability question. In California, claims over non-copyrighted content were already dismissed with prejudice, so Google was required to allege content that qualifies. A further footnote notes that the New York court, in accepting Reddit's user-authorisation allegations, relied on Reddit's copyright licence and Public Content Policy while separately rejecting reliance on a user privacy provision, on the reasoning that a promise to protect personal information does not establish authority over copyrighted content.

SerpApi is also litigating as plaintiff, pursuing a trade secret case against rival SearchApi in the Western District of Texas, where the defendants moved to dismiss on April 20, 2026.

Why this matters for marketers

The commercial question underneath the pleading dispute has not changed since December. Rank trackers, competitive intelligence dashboards, keyword research platforms and share-of-voice tools all read search results pages, whether directly or through vendors that do it for them. If an anti-bot layer over a public results page qualifies as a copyright access control, statutory exposure of $200 to $2,500 per act attaches to reading a page, and it scales with query volume rather than with the size of the company doing the reading.

The July order answered that question in the negative for the general case, and answered it permanently: claims covering non-copyrighted search results cannot be refiled. What survives is confined to specific licensed works, which is a narrower surface than the December complaint described. The removal of the Google Shopping and Google Maps theories narrows it further, because merchant product imagery and business photography are precisely the categories retail media and local search tools touch most often.

Access economics have been tightening independently of the litigation. Google eliminated the num=100 parameter on September 14, 2025, forcing platforms to make ten requests where one had sufficed. Semrush confirmed the tenfold cost increase within a day. Ahrefs restored top-100 tracking at the end of October 2025 through unspecified means. SerpApi's own Light Fast API workaround was restricted to three organic results in early October 2025. Litigation is one instrument among several.

The September 29 hearing will determine whether Google's second attempt clears the pleading bar, and whether a third attempt is available. SerpApi has asked for dismissal with prejudice.

Timeline

Summary

Who: SerpApi LLC, the Austin, Texas search-data provider founded in 2017 by Julien Khaleghy, represented by Colin R. Kass, David A. Munkittrick and Joseph E. Clark of Proskauer Rose and Heather Tovar of Ad Astra Law Group, filed against Google LLC, represented by David H. Kramer of Wilson Sonsini Goodrich and Rosati. Chief Judge Yvonne Gonzalez Rogers presides. SerpApi General Counsel Chad Anson issued the company's on-the-record statement.

What: A 23-page motion under Rule 12(b)(6) asking the court to dismiss Google's amended DMCA complaint with prejudice, on the ground that Google has not plausibly alleged copyright owner authorisation to implement SearchGuard through any of its four claimed sources: two unidentified licence agreements, the Reddit agreement and Google's Privacy Policy. The motion argues that three of those agreements were never placed before the court or quoted, that a download restriction is not an access restriction, and that the Privacy Policy governs user data rather than copyright. Five previously rejected grounds are renewed for appellate preservation.

When: The motion was filed on August 24, 2026, as Document 49. Google's amended complaint was filed on August 10, 2026, following the July 20, 2026 dismissal. A hearing is set for September 29, 2026 at 2:00 p.m.

Where: The United States District Court for the Northern District of California, Case No. 4:25-cv-10826-YGR, Courtroom 1 on the fourth floor. The parallel Reddit litigation continues in the Southern District of New York, and SerpApi's trade secret case against SearchApi continues in the Western District of Texas.

Why: The July order held that an anti-bot layer over non-copyrighted search results falls outside the DMCA, permanently. What remains turns on whether Google can show that the owners of specific licensed works authorised the access control. The answer determines the statutory exposure attached to reading a search results page, which matters to every rank tracking, competitive intelligence and keyword platform that depends on that data.