A few months ago this Kat pondered the shifting boundaries of the web following pending *Reddit v. Perplexity *litigation (IPKat here). As the scramble for data intensifies, the ‘OG’ internet infrastructure provider, Google, has finally ‘entered the chat’, indicating that we have truly reached a turning point regarding access to an already limited ‘open’ web. Perhaps a Christmas miracle for some, Google accused SerpApi days in reach of holiday break of copyright infringement. Strikingly similar to Reddit’s lawsuit, a Google partner, the lawsuit centres on the circumvention of technological protection measures (TPM), including both the act and the dissemination of a circumvention device (section 1201 (a)(1A) and (2)(A) US Copyright Act).

SerpApi’s business model, described as ‘parasitic’ in the complaint, turns ‘Google’s 10 blue links into a business’. SerpApi offers access to search engine result pages (SERPs) through its own application programming interface (API). The business model is literally the name as SerpApi uses an automated process to access Google’s SERP ‘en masse’ for third parties: **advertised **as ‘Scrape Google AI Mode’ starting at $75 per month. Testing Google’s reCAPTCHA and Cloud Armour, even Google’s latest internal security measure, SearchGuard, launched in January 2025, was ultimately defeated by SerpApi ‘creating fake browsers using a multitude of IP addresses that Google sees as normal users’ to access and scrape content without authorisation.

The complaint centre’s on Google’s ‘corporate mission’ to ‘organize the world’s information and make it universally accessible and useful’. Google Search locates information, abiding opt-out mechanisms (e.g. robot.txt.protocol), and uses algorithms to ‘identify and rank results based on Google’s views as to what information the user may find the most relevant’. This is an iterative process that involves ‘decades of work and billions of dollars to develop and operate Google Search, and the service processes billions of user queries a day’.

SerpApi endangers Google’s licensed partners for services related to Google Search (e.g. Knowledge Panels, Google Maps and Google Shopping). Relationships that SerpApi allegedly puts at risk because it deceptively accesses this content automatically at ‘astonishing scale’ and ‘denies Google’s partners compensation for their works’. SerpApi also fails to compensate the costs that their automated processes impose on Google’s ‘computer infrastructures’. What emerges is an argument centred upon harmful misappropriation.

**A public relations feat: from the open web to a TPM believer **

Critics have been quick to **remind **us that Google became a tech giant through crawling and caching the web. They benefited from open access concepts that TPM-related infringement has historically compromised (e.g. video games and streaming). Field v. Google in the US, or the Meltwater saga in the UK and EU, delineated the boundaries of scraping publicly available data, namely through opt-out mechanisms, implied permissions and narrowly-defined exceptions. As you can imagine, residing on the other side of the copyright fence would seem a difficult pivot for Google.

Yet, their public relations team has managed it almost seamlessly. They cast Google as fighting ‘the good fight’ on behalf of copyright owners whose content appears in Google search and is allegedly scraped without consent by SerpApi. They lean on SerpApi’s bad faith actor characterisation to argue that SerpApi violates ‘service governance agreements’ and ‘flouts access restrictions that those services convey to automated crawlers or “bots” through robots.txt. instructions’.

As Google does not offer an API for its search results, its SERPs hold incredible value for AI chatbots like Perplexity. The question is how to frame the value attached to indexing the web, unsurprisingly, the answer is somewhat Lockean: investment and labour. Google claims that SerpApi ‘appropriates the output of other services that have made substantial investments to generate it, and delivers the content to third parties for a fee’. Google’s rhetoric is almost convincing. Related rights have long protected the financial, technical and organizational investments related to a copyright work. Extended further, protection against the circumvention of TPMs, a notorious form of para-copyright, ensures the commercial exploitation of a work given the internet’s ability to exponentially copy and share content. Still the picture is incredibly muddled when it comes to AI-supported search and chatbots.

**The race to protect data access, infrastructure and compute capacity **

Nearly two years ago, Google and Reddit entered into a new **$60 million per year **agreement that provides access to Reddit’s Data API that ‘delivers real-time, structured, unique content from their large and dynamic platform’ and ‘enables Reddit to integrate new AI-powered capabilities using Vertex AI’. As this Kat sees it, this lawsuit is about protecting Google’s market dominance alongside the longstanding copyright question: access, and now, by an extension of investment logic, infrastructure and compute capacity.

Google argues that it must invest more to protect Google Search content and differentiate between legitimate users and bots without downgrading the experience of ordinary users. More generally, SerpApi’s activity challenges online business models based on gatekeeping internet traffic and monetization such as Cloudflare. Unsurprisingly, they view companies like Perplexity that use SerpAPI to evade no-crawl directives as bad faith actors. In response, Cloudflare now** offers, in closed beta, a ‘pay per crawl’ initiative where ‘publishers, content creators and website owners’ can charge AI crawlers **to monetise AI access.

The question that looms in the background is who pays to technically support the automated requests made to Google, and third parties like Cloudflare, that scale far beyond human capability? According to Google, it is an increase of 25,000% over a two-year period, with others explaining that ‘the cost per query of an AI chatbot can be as much as 10 times the cost of a conventional Google search’. Ironically, Google’s AI tools like Gemini also detrimentally **impact **internet traffic for news publishers and content creators. While Google operates a globally-scaled infrastructure, leaked slides from a Google AI Infrastructure meeting show that Google intends to ‘double every 6 months [compute capacity] […] the next 1000x in 4-5 years’ to meet demand for AI services.

**Comment **

Without a solution in sight, we are inching towards an incredibly fragmented online future where search results are possibly linked to whether an AI-driven search engine pays to crawl. As problems arise over the selection of ‘valuable’ data to scrape, what we see and who displays it is likely to be increasingly limited. This licensing environment invariably raises the cost for third-parties to access high-quality search data. It means using Google is simply unavoidable.

This reality has not escaped regulators. The UK has confirmed that Google has strategic market status in search services. The EU has opened an investigation into possible anticompetitive conduct by Google in the use of online content for AI purposes. Last year, in United States v. Google a federal court ruled that Google is required to share to ‘Qualified Competitors’ certain search-index and user-interaction data to ‘help improve the quality of rival search engines and mitigate the consequences of Google’s illegal conduct’.

Google contends that ‘SerpApi’s fakery takes many forms’, and this Kat suggests that fakery also extends to using the copyright system to create a ‘search-data lock-right’. Framing SerpApi’s activity as TPM infringement echoes concerns related to opt-out mechanisms (e.g. article 4 of the CDSM) that imply infringement of the reproduction right. As evidenced already in Stability AI v Getty Images (IPKat here), mapping AI models on to the copyright infringement is complicated. Conversely TPM infringement presents a swifter and sharper tool to suppress unauthorized access and large-scale extraction of online content without evaluating whether the use infringes the reproduction right. Such an approach risks entrenching a model of access control that ultimately reshapes who can see, reuse and build on the open web at all.

Fakery, free-riding and the fight for SERP data for AI-related search and chatbots Reviewed by Georgia Jenkins on Tuesday, January 06, 2026 Rating: 5