The Google SerpApi lawsuit over AI scraping has run into a basic copyright problem: Google does not own the search-result content it says SerpApi scraped. A court last week granted SerpApi’s early motion to dismiss, finding that Google had not shown it had standing to sue under the Digital Millennium Copyright Act.
Google sued SerpApi in December, accusing the web-scraping company of getting around Google’s anti-scraping systems and selling data pulled from search results through an unauthorized “Google Search API” service. Google said those controls were meant to protect copyrighted material that can appear in results, including content licensed for knowledge panels.
The legal hook was the DMCA’s anti-circumvention rules, which are usually aimed at bypassing technical measures that control access to copyrighted works. Search results themselves are not copyrightable, and the judge found Google had not alleged enough to show it owned the protected works or was acting for rights holders.
Why was the Google SerpApi lawsuit dismissed?
The court said Google lacked DMCA standing because it did not own the content in the search results and had not shown that rights holders authorized Google to sue on their behalf. Meredith Rose, senior policy counsel at Public Knowledge, told Ars Technica that the ruling turned on Google’s failure to identify what copyrighted material it was protecting.
Rose said Google and Reddit appear to be using whatever legal tools are available as AI-related scraping has surged over the past three years. She described their use of the DMCA in this context as unusual, while noting that the law has long been used to stop unwanted content uses and force licensing discussions.
Google is not done. Spokesperson José Castañeda told Ars Technica that Google plans to amend its complaint and said the court rejected most of SerpApi’s other arguments about standing. The court gave Google 21 days to file an amended complaint.
That amended version will likely have to thread a narrow needle. Google has acknowledged that search results cannot be copyrighted, while arguing that knowledge panels sometimes include licensed copyrighted material. Rose told Ars Technica that Google could create new risk for itself if it argues too broadly that knowledge panels are filled with copyrighted material it reproduces.
How does Reddit fit into the scraping fight?
Reddit filed a similar case in October against SerpApi and Perplexity, accusing them of scraping Reddit material that appears in Google results. Reddit argued that SerpApi evaded protections on Reddit’s own platform and Google’s controls around Reddit content in search results.
Reddit did not respond to Ars Technica’s request for comment, but said in a filing before a hearing that it was prepared to address how Google’s court loss affected its own case. SerpApi said Reddit was not present in the courtroom and said the judge appeared focused on whether Reddit’s agreement with Google authorized Google to protect Reddit’s copyrighted content.
Rose told Ars Technica that the Google ruling may be a bad sign for Reddit because Reddit cannot claim to be the owner or exclusive licensee of content displayed in search results. She said the court in Google’s case identified three possible paths for DMCA standing: copyright owner, exclusive licensee, or the party deploying and making the technological protection measure at issue.
SerpApi told Ars Technica that Google and Reddit are trying to use the DMCA to fence off public web data they did not create or own. The company said its customers, including Nvidia, Uber and Adobe, use SerpApi for structured access to search data, and that the lawsuits have created uncertainty despite continued business growth.
The broader fight is over who gets to automate access to public web information as publishers and platforms try to limit high-volume scraping. Rose warned that broad anti-scraping reactions can also harm research, archiving, journalism and public health work that depend on automated crawling at scale.
This story draws on original reporting from Ars Technica.
