Google's DMCA Case Against SerpApi Was Just Dismissed. Here's Why It Matters for Scraping

Disclaimer: This is written by engineers who build a data API, not by lawyers. It's a plain-language read of a ruling that dropped on July 20, 2026. It is not legal advice. Talk to an actual attorney about your specific situation.

For most of the last decade, the scariest weapon a big platform could point at a scraper was the CFAA, the anti-hacking law that hiQ v. LinkedIn eventually defanged for public data. So when Google sued SerpApi in December 2025, it reached for a different, arguably more dangerous tool: the DMCA's anti-circumvention rules. The theory was basically "our anti-bot system is a lock, you picked it, that's illegal" — the same part of copyright law written to stop people from ripping DVDs.

On July 20, 2026, that theory fell apart. Chief U.S. District Judge Yvonne Gonzalez Rogers of the Northern District of California granted SerpApi's motion to dismiss and threw out both of Google's DMCA claims. Part of the case is dead permanently. A narrow sliver survives on life support. Either way, the core argument — that blocking bots from public search results is, by itself, copyright circumvention — did not survive contact with a judge.

If you build anything on public web data, this one is worth understanding properly, not just from the headline.

What Google actually argued

Google filed on December 19, 2025, with two claims under Section 1201 of the DMCA:

• Section 1201(a)(1)(A) — that SerpApi circumvented "SearchGuard," Google's anti-scraping/anti-bot system, to reach search results.
• Section 1201(a)(2) — that SerpApi trafficked in that circumvention by selling a service built around it.

SerpApi, if you haven't used it, sells programmatic access to Google search results through an API. That output feeds a huge amount of the SEO world: rank trackers, keyword tools, competitive-intelligence dashboards. So Google's framing was that SerpApi bypassed SearchGuard, scraped results at scale, and resold them — imposing costs on Google and undercutting its own licensing deals.

The catch, and the whole case turns on this, is that Section 1201 doesn't protect "systems." It protects access controls that guard a work protected under copyright. Google had to show SearchGuard was locking a copyrighted thing. That's where it ran into trouble with its own complaint.

The split ruling, and why the split is the whole story

Judge Gonzalez Rogers didn't hand down one verdict. She split Google's claims into two buckets and treated them very differently.

Bucket one: search results with no copyrighted content. This is the ordinary case, a results page of links, snippets, and factual index data pulled from across the web. The court dismissed both claims here without leave to amend, meaning Google can't refile them. The reasoning is clean: where results carry no copyrighted content, SearchGuard, in the court's words, "cannot be said to effectively control access to a work protected under the Copyright Act." No protected work, no Section 1201 hook. Plain, aggregated search results, URLs, snippets, factual data, are publicly accessible facts, and facts aren't copyrightable. That claim fails as a matter of law, so amending it would be pointless.

Bucket two: results with a copyrighted component. Sometimes a results page shows a Knowledge Panel that "may contain" a licensed image or other copyrighted material. For that narrow situation, the court dismissed the claims with leave to amend, giving Google 21 days to try again. But it flagged a second, independent problem: to assert anti-circumvention protection over third-party copyrighted material, Google needed to show the actual copyright owners authorized it to deploy SearchGuard on their behalf. Section 1201(a)(3)(B) requires the measure to operate "with the authority of the copyright owner." Google didn't allege that. So even the surviving sliver has a hole in it Google now has to plug.

Read those two buckets together and the picture is clear. The part Google can never refile is the part that covers how scraping actually works day to day: pulling organized public listings off a results page. The part it might revive is confined to the odd case where a licensed image happens to sit in a panel on the page.

It wasn't a clean sweep for SerpApi

Here's the honest bit that the celebration tweets skip. SerpApi also argued Google had no right to sue at all, on the theory that the DMCA only protects copyright owners, and Google doesn't own the third-party results it indexes. The judge rejected that. She held that Section 1201's protections aren't limited only to copyright owners, so Google had standing to bring the claims even if it wasn't going to win them.

And the court did not say "scraping is legal, go wild." It accepted, at least in principle, that deceptive evasion of an access control could be unlawful circumvention under the DMCA. What killed Google's case wasn't that anti-circumvention law can never touch scrapers, it's that Google couldn't tie SearchGuard to a protected work the way the statute demands. That's a pleading-and-copyright problem, not a blanket blessing for bots.

SerpApi's CEO, Julien Khaleghy, framed the outcome as a win beyond his own company, calling it a win for "all who depend on an open internet" (Search Engine Land). Google, as of the rulings being reported, hadn't commented.

Why this is genuinely a win for the scraping community

Strip away the procedure and the takeaway is one most of us have argued for years, now reinforced by a federal court: you can't use copyright law to fence off facts.

A list of URLs, a snippet, a follower count, a price, a public post's like total — these are publicly visible facts. The DMCA was built to stop people from breaking the encryption on a copyrighted movie or ebook, not to convert every anti-bot script into a copyright violation. Google tried to stretch it to cover the second thing, and the court declined. That matters because Section 1201 is, in some ways, scarier than the CFAA: it doesn't require proving unauthorized access to a computer, just circumvention of a "technological measure." If that theory had worked, any site could slap a bot-detector on public pages and reclassify ordinary scraping as a federal copyright offense. This ruling closes that door for plain public data.

It also lines up neatly with the public-vs-private line that has defined this whole area since hiQ. If you want the fuller history, we walked through every major web scraping court case and the 2026 legal landscape for public vs. private social data separately. The pattern holds: public, unauthenticated data sits on very different legal ground than anything behind a login.

The limits (read this before you screenshot the headline)

I'd be breaking our own rule about honesty if I let you walk away thinking this settled everything. It didn't.

• It's a motion-to-dismiss ruling, not a final judgment. The court tested whether Google's complaint stated a valid claim, and found it didn't. It's a strong signal, not the last word.
• Google still has 21 days to amend the Knowledge Panel portion. If it comes back with the copyright-owner authorization the court said was missing, that narrow claim could continue. Discovery is paused until then.
• This is about the DMCA specifically. It doesn't touch the CFAA, breach-of-contract/terms-of-service claims, trespass to chattels, or data-privacy law. A platform can lose a Section 1201 argument and still come after you on a different theory.
• Copyrighted elements are still copyrighted. Scraping plain listings is one thing; copying licensed images, full articles, or other original creative work is a different, riskier thing.
• This is one district court. It's persuasive and important, but it isn't binding nationwide, and Google may appeal or refile.

There's also a reminder that the fight isn't over industry-wide: SerpApi is separately facing a DMCA suit from Reddit raising similar questions about scraping publicly viewable pages. So the theory Google fumbled here will get tested again elsewhere.

What it means for how we think about data at SociaVault

We've always drawn our line in the same place this ruling reinforces. SociaVault only collects publicly available data — public profiles, public posts, public comments, public engagement metrics. We don't log in, we don't touch private accounts, DMs, or protected content, and we don't try to pass off copyrighted media as free-to-take. That stance isn't a reaction to this case; it's in our compliance statement and it's why we treat customers as the data controller responsible for how they use what they pull.

Rulings like this one don't change what we do. They just make the ground under public-data tooling a little more solid. The internet's whole premise is that public information stays usable — for developers, researchers, and the AI companies now leaning hard on it. A court just declined to let one of the biggest platforms narrow that premise with copyright law.

Frequently Asked Questions

Did the court rule that scraping Google is legal?

Not exactly. It ruled that Google's specific DMCA anti-circumvention claims failed, because plain search results aren't copyrighted works and Google didn't show it had copyright owners' authorization for the parts that might be. The court left open that deceptive circumvention of access controls could, in other circumstances, violate the DMCA. So it's a narrow, important win, not a blanket "scraping is legal" declaration.

What is Section 1201 of the DMCA, in plain terms?

It's the "anti-circumvention" part of copyright law. It makes it illegal to bypass a technological measure that controls access to a copyrighted work, think the encryption on a DVD or a paywall around a copyrighted article. The key requirement is that there has to be a copyrighted work behind the lock. That's the requirement Google couldn't satisfy for ordinary search results.

Why did part of the case survive?

Search results sometimes include a Knowledge Panel that may contain licensed images or other copyrighted material. For that narrow slice, the judge gave Google 21 days to amend and try again, but only if it can allege that the actual copyright owners authorized Google to protect that material with SearchGuard.

Does this apply to scraping social media data?

Not directly, it's a Google search case, but the underlying principle is the same one that governs social data: public, unauthenticated facts sit on much stronger legal ground than content behind a login or original copyrighted media. We cover that distinction in our public vs. private data guide and our Instagram scraping legality breakdown.

Can Google appeal or refile?

It can amend the Knowledge Panel claim within 21 days, and it retains normal appeal rights. The portion dismissed "without leave to amend" is the part it cannot simply refile. Expect the broader question to resurface, SerpApi is also defending a similar DMCA suit brought by Reddit.

Is this binding on other courts?

No. It's a decision from one federal district court (Northern District of California). It's influential and well-reasoned, and courts do look to each other, but it isn't nationwide precedent the way a Supreme Court ruling would be.

---

If you're building on public data and want an API that treats the public-versus-private line as a hard rule rather than a suggestion, that's exactly how we built SociaVault. You can start free with 50 credits, no card required, and see what clean, structured public data looks like across 25+ platforms.

Sources: Search Engine Journal, Search Engine Land, PPC Land, and Search Engine Roundtable. Content was rephrased and summarized for compliance with licensing restrictions.