William del Strother, VP of Innovation, Friend MTS
For two decades, the anti-piracy industry has operated under a quiet assumption: keep infringing content out of Google's results, and you have cut off the lion's share of demand. That assumption was always a simplification. In 2026, it has become genuinely dangerous.
The discovery layer of the internet has evolved. AI-driven answer engines, regional search incumbents, and social super-apps have all become primary routes through which viewers find content — and the rules by which a pirate's URL reaches a viewer have changed faster than most rights-holder strategies have kept pace with.
At Friend MTS, we have been closely tracking this shift. The picture that emerges raises an uncomfortable question: is the industry protecting the right doors?
The scale of the change
Google still processes roughly 16.4 billion searches a day and holds around 90% of traditional search share globally. But almost 60% of those searches now end without a click — the answer is rendered directly on the result page through AI Overviews and Knowledge Panels.
At the same time, AI-native tools such as ChatGPT and Perplexity are growing at more than 200% year-on-year. ChatGPT alone handles an estimated 2.5 billion search-style queries a day. Whilst some AI tools have guardrails that refuse to serve results for pirated content, some will still deliver results for illegal streams.
This is not a story about one piracy gateway being replaced by another. It is a story about that gateway multiplying.
Nine indexes, nine different piracy economies
Each major AI assistant is, in effect, looking at a different web. ChatGPT relies primarily on Bing's index. Claude has standardised on Brave Search — an independent index of more than 30 billion pages. Perplexity built its own crawler. Grok pipes directly into X (formerly Twitter) and indexes tweets instantly, allowing users to locate streams before de-listing has a chance to take effect.
The consequence is straightforward: the same query run through five different AI tools returns five materially different sets of citations. A single infringing site can be invisible in one assistant's recommendations and prominently cited in another. Delisting from Google does not remove it from Brave's index. Action on Bing does not touch Perplexity's proprietary crawl.
Layered on top of that fragmentation is a regional and social dimension that Western-headquartered rights-holders frequently underestimate. Baidu processes around six billion searches a day in China, where Google sits at just 2% share. Yandex dominates more than 70% of Russian users. Naver holds roughly half of South Korea. And none of those figures captures the social super-apps: Xiaohongshu, Douyin, and WeChat search are, for hundreds of millions of users, the search bar that matters.
For the first time in the modern internet, content protection has to be planned against a plural discovery layer. An anti-piracy programme that covers Google and Bing in 2026 is still, on a generous reading, leaving half the world's discovery traffic unmonitored.
What this means for anti-piracy
The fundamentals of piracy have not changed: operators, hosting, monetisation, and a viewer at the end. What has changed is the visibility layer — and visibility is ultimately the lever that determines how much of the pirate economy is commercially viable. There are key questions to answer to ensure anti-piracy services continue to be effective.
How does notice-and-takedown translate to nine concurrent indexes, some opaque, some operated by entities with no relationship to the rights-holder community? What is the equivalent of a Google delisting for Brave's surface or Perplexity's curated crawl?
How do we measure exposure when search rank is no longer a single number? A pirate site invisible on Google may still be one of three citations in a ChatGPT answer.
How do we use AI on the defending side as fast as pirates use it on the offending side? Generative tools lower the cost of producing multilingual piracy infrastructure at scale. Deployed well, they lower the cost of detection at the same speed.
How Friend MTS is responding
These questions are shaping our detection architecture at Friend MTS and the services we are building for broadcasters, sports rights-holders, and studios.
Our content monitoring services already operate across a fragmented discovery environment — tracking infringing streams across platforms and regions that a Google-only approach would miss. This is powered by fingerprint-augmented content monitoring, AI-driven pattern recognition, and forensic watermarking working together to identify illegal streams wherever they surface, spot patterns across thousands of new domains a week, and trace a leak back to its source, not just its distribution.
We're extending this further with semantic matching across regional languages and automated evidence packaging across multiple retrieval surfaces simultaneously. Detection without enforcement is intelligence without value — so we keep investing in both sides of that picture.
The rights-holders who win the next decade of anti-piracy will be the ones who built coverage across this fragmented map before it consolidated, not after.
The road ahead
The strategic picture is not yet settled. We do not know which AI assistant will become the default discovery surface in five years, or whether regulators will extend search obligations to AI retrieval. But we are confident that the window to act is now, while the landscape is still forming.
Anti-piracy has always been a race between detection and distribution. The track just got wider. Friend MTS intends to run on all of it.
If you want to understand what that means for your content protection programme, contact us today.
