I wrote a post almost a year ago titled “AI Search Will Break the Internet.” Well, that day has come. AI search has officially broken the Internet. Some of the first quantitative analyses of referral traffic from AI search tools have been released, and the numbers confirm the fears.
Tollbit State of the Bots Report
Let’s start with Tollbit’s State of the Bots report.
This is a bleak story for content creators. Bots from AI search tools are scraping content without permission and sending virtually no traffic back to the content creators.
Note that traditional search tools like Google abide by robots.txt files (that websites create to specify what can be scraped) and then send high volumes of traffic that drive ad revenue or sales. Now AI search tools are neither following the rules nor sending revenue-generating traffic.
ChatGPT Referrals
And here’s another report from Ezra Eeman confirming the same trends, specifically for ChatGPT.
Google in Trouble
These trends aren’t only disruptive to publishers but also to search engine incumbents like Google. To stay relevant, Google has introduced more AI search functionality into their main search engine, which inevitably causes drastically lower click through rates to publishers. The impact was predictable:
This is further stretching the already-questionable value exchange between content creators and Google. Regulators are investigating this as a potential abuse of a monopoly position.
David Buttle
In my original post on this topic from last year, I argued for legal action against Google. So no surprise from me here.
Reshaping the Internet
The ecosystem of content creators and search engine referrals that has underpinned the free and open Internet since its early days is on the verge of collapse.
Here’s how I see the future of the Internet shaping up.
1. AI as Primary Tool for Information Retrieval
AI search is just a better experience for users than traditional search. Users will continue migrating more and more of their search queries away from Google and towards AI chatbots with Internet connections (e.g. Perplexity, ChatGPT, etc.).
2. Data as a Service for News Publishers
News publishers should expect fewer eyeballs on their websites, but not fewer readers of their insights. Publishers need to adopt Data as a Service (DaaS) revenue models that either meter bot traffic (e.g. via Tollbit) or deliver content via API or MCP.
3. Content Marketing in the AI Age
Content marketing (e.g. blogs that inform customers during their buying journey) has to adapt. The goal of this content is to lead to a buying decision. In contrast, the goal of news data is to provide facts.
When an AI search engine answers a user query with a fact sourced from a news article, the value of that fact has already been delivered to the user. So news publishers need to capture value before the fact is delivered to the user (via metering).
But content marketing is different. A content marketer wants the AI search engine to use their content to help inform the user during some buying decision. The value is in the purchase itself, not the data extracted from the content. Metering is not useful here; in fact, the easier the content is to find the better.
Brands should want their content scraped by AI bots and freely delivered to users. Even better if your content ends up in the training set of the AI – then they don’t even need to scrape your website for each user query.
3.1 Developer Relations
I want to add a special note about content marketing strategy for developer tools (software or services used by developers to build things). In the past, StackOverflow was a place where developers could get help using tools and packages by asking the community a question. They almost always had the first hit in Google for any given technical question because the community was thriving, generating answers to millions of questions.
But now developers just ask ChatGPT or Cursor. Which means no one is creating free content for your developer tool anymore on StackOverflow. Creators of developer tools need to find new ways to create technical content for their tools, at scale, so that ChatGPT and Cursor can provide answers to questions about your tool.
Most developers don’t have time to write extensive tutorials on every possible use case for their tools. And no one is doing it for them for free on StackOverflow anymore. I anticipate that creating incentives for other people to write content about your tool will be an important strategy in the AI age. I’ve created an AI Blogathon here on Skillenai to facilitate just that.
In the past, search engine traffic was the primary source of traffic for most websites. With abysmal referral rates from AI search tools, I would expect that social media traffic becomes the primary source in the future. So a strong social presence for both publishers and brands will be more important than ever.
The Internet Has Been Reshaped—Now What?
AI search has already broken the Internet as we knew it. The foundational bargain between content creators and search engines—free content in exchange for traffic—is collapsing. Instead, AI search tools are extracting value while returning little to nothing to publishers.
The future belongs to those who adapt. Content creators must rethink their monetization models, brands must reshape their content marketing strategies, and developers must find new ways to educate users about their tools. Social media will likely become the dominant traffic driver, and new economic models like Data as a Service will emerge to counterbalance the AI disruption.
The free and open Internet isn’t disappearing, but it is being fundamentally reshaped. Those who recognize this shift and adjust their strategies now will be the ones who thrive in the AI-driven future.
