The whole digital publishing world got a nasty surprise in early 2026. For news outfits like Press Gazette, the threat of AI scrapers wasn’t some academic discussion. It was a direct hit on their business model. Their entire revenue stream, built on selling access to proprietary content, was being eaten away as large language models (LLMs) hoovered up their articles with zero compensation. This was an existential threat to their intellectual property, a fight for survival. How was a traditional publisher supposed to defend itself when the internet itself was learning to distill information through algorithms that often forgot to mention where they got it from, like Press Gazette?
Key Takeaways
- Publishers like Press Gazette are fighting back with technical measures, deploying IP blockers and trying to obfuscate content to stop unauthorized AI scrapers.
- Brave browser’s “Unscrapable” feature gives publishers a client-side weapon, scrambling webpage data before an AI model can use it for training.
- The European Union’s AI Act, which kicked in in 2025, gives publishers legal teeth by demanding transparency from AI models about their copyrighted training data.
- A real defense strategy has to be multi-pronged, mixing technical blocking with legal action under new laws and, frankly, adapting the business model itself to protect content.
- If AI scraping goes unchecked, the economic hit could be huge, with some publishers already projecting 15-25% drops in referral traffic, which directly impacts the money available to produce quality journalism.
The problem at Press Gazette, a publication known for its sharp analysis of the media industry, became impossible to ignore when their analytics lit up with bot traffic from weird IP ranges. At the same time, their direct organic search traffic for specific, hard-to-report stories was tanking. “We noticed our unique reporting, often requiring extensive journalistic effort, was appearing verbatim or heavily summarized in AI-generated responses without any referral back to us,” explained Sarah Jenkins, Press Gazette’s Head of Digital Strategy, in a recent industry webinar. “It felt like our content was being stolen in plain sight, then regurgitated by machines that offered no link, no subscription prompt, nothing.”
AI Scraping: Devaluing Content in Real Time
Publishers have been fighting ad blockers and aggregators for years, but AI scraping was a different beast entirely. These weren’t just simple bots indexing for a search engine. They were algorithms designed to understand, process, and then re-package information so completely that the original source became irrelevant. The money side of this was grim. As a Reuters report from late 2025 pointed out, news publishers globally were already bracing for a 15% to 25% drop in referral traffic because generative AI was just answering questions directly. That meant fewer page views, which meant less ad revenue and fewer chances to turn a reader into a paying subscriber.
Initially, Press Gazette’s tech team, led by Mark Harrison, threw the standard playbook at the problem. They tightened up their robots.txt file to disallow known AI crawlers and implemented more aggressive IP blocking. “We even tried some JavaScript-based obfuscation,” Harrison recounted, “rendering parts of the content dynamically to make it harder for static scrapers to grab. But it was a cat-and-mouse game. The AI models, backed by immense computing power, simply adapted.” The scrapers got smarter, learning to mimic human browsing by rotating through thousands of IP addresses and even executing JavaScript to see the final, rendered page. This was a battle to preserve the actual value of their journalism.
Brave’s “Unscrapable” Feature: A Client-Side Countermeasure
Press Gazette’s strategy shifted when they started looking beyond their own servers and at client-side solutions. This search led them to a partnership with Brave browser, which had just launched its “Unscrapable” feature, a tool developed specifically to fight this kind of data extraction.
The “Unscrapable” feature works by messing with the webpage’s structure inside the user’s browser *before* any AI or scraper can get its hands on the clean data. It injects invisible junk data, shuffles element IDs, and even adds random rendering delays, all of which is unnoticeable to a human but makes the page a chaotic mess for an automated scraper. “It’s like serving a human a perfectly legible page, but giving a robot a page full of gibberish,” explained Ben Smith, a senior engineer at Brave, at a joint press conference. The whole point is to make scraping economically unviable by ensuring the cost of cleaning up the scrambled data is far higher than the value of the data itself.
Press Gazette’s team integrated Brave’s API into their content delivery network (CDN) configuration, which gave them granular control. They could now selectively turn on “Unscrapable” mode for certain types of content or for traffic that looked like it was coming from a bot. “We didn’t want to break the user experience for legitimate readers,” Jenkins emphasized. “The implementation needed to be surgical, targeting the scrapers without affecting our loyal audience.” They started by protecting their premium, subscriber-only articles and their most in-depth analytical pieces, the exact kind of content that LLMs love to summarize.
Early Results and Industry Reaction
Within three months, Press Gazette saw a real difference. Harrison’s team reported a major drop in suspicious bot activity on the protected parts of the site. This correlated with a stabilization, and in a few cases, a small rebound, in direct traffic to those same articles. Just as importantly, they saw fewer instances of their unique analysis showing up in AI-generated summaries without a link or credit. “It wasn’t a silver bullet, nothing ever is,” Harrison admitted, “but it definitely raised the barrier for entry for these scrapers. The signal-to-noise ratio for them became too low.”
The rest of the publishing industry was watching closely. The threat of AI scraping isn’t unique to Press Gazette. As The Financial Times reported in late 2025, nearly every major news organization was looking at similar technical fixes and legal options. The European Union’s AI Act, which went into full effect in early 2025, suddenly became very important. Specifically, Article 52 of the Act gives publishers a real legal weapon, forcing providers of general-purpose AI models to publish a detailed summary of the content used for training and to respect EU copyright law. This gives publishers a solid legal basis to challenge the unauthorized use of their content.
The technical arms race, however, continues. Some experts, like Dr. Anya Sharma at the University of Cambridge, argue that these client-side solutions are in the end just reactive. “While tools like Brave’s ‘Unscrapable’ are clever,” she noted, “the underlying problem is the ease with which AI models can be trained on vast datasets without clear consent or compensation. We need a fundamental shift in how AI developers approach data acquisition, rather than relying solely on defensive measures.” Her research, published in the ACM Journal of Computers and Society, pushes for a “permissioned web” where content owners can grant or deny AI access and get paid for it.
The Path Forward: A Multi-faceted Defense
The experience with Brave taught Press Gazette that they had to fight on multiple fronts. They’re still refining their technical blocks and updating their scraper detection algorithms. At the same time, their legal team is ready to fire off cease and desist orders based on the EU AI Act. As Jenkins asserted, “It’s about establishing precedent. If AI companies know there will be legal consequences, like a letter citing the new law, they’ll be more inclined to negotiate licensing agreements.”
Beyond pure defense, Press Gazette’s internal conversation also shifted to adaptation. Their goal now is to create content that’s just more valuable to a human than to a machine, focusing on deep investigative work, building stronger community features, and producing analysis that’s too nuanced for an AI to properly replicate or synthesize. They’re also looking at new revenue streams that are less vulnerable to this kind of disruption, like running exclusive events, selling specialized data services, and pushing direct reader support. The fight isn’t just about protecting what they have. It’s about redefining the value of human journalism in a world full of AI.
The Press Gazette-Brave partnership shows one way publishers are fighting back, proving that while technology creates problems, it can also provide some of the tools to solve them. It’s forcing content creators to think harder about their digital assets and their direct relationship with their audience. This kind of constant adaptation is probably the future of journalism.
What happened with Press Gazette and Brave makes one thing clear for anyone creating content online: a passive defense against AI scraping isn’t enough anymore. It’s now essential to implement technical safeguards, understand the new legal ground rules, and constantly re-evaluate what makes your content uniquely valuable in a world increasingly shaped by artificial intelligence.
What is AI scraping?
It’s the automated process where AI models crawl websites and other online sources to extract massive amounts of data. This data gets used to train the AI to generate its own content or answer questions, usually without asking permission from or paying the original creators.
How does Brave’s “Unscrapable” feature work?
It’s a client-side tool that messes up a webpage’s code and structure inside the user’s browser before a scraper can see it. By injecting junk data and shuffling elements, it makes the information worthless for AI training while keeping it perfectly readable for a human.
What are the main challenges publishers face from AI scraping?
The big ones are a major loss of referral traffic, which hits ad revenue hard, and the general devaluation of their original work. It also means their copyrighted material is being used without permission to train a competing product, which undermines their entire business model.
Are there legal protections against AI scraping?
Yes, and they’re getting stronger. The EU’s AI Act from 2025, for example, forces AI companies to be transparent about their training data and respect copyright. This gives publishers a legal tool to go after unauthorized use, backing up their technical defenses.
What strategies can content creators use to protect their work from AI?
You need a layered approach. Use technical tools like better robots.txt files, IP blocking, and client-side obfuscation. Get familiar with the new legal options available to you. And, most importantly, adapt your business to focus on high-value content and direct reader engagement that an AI can’t easily replace.
