We are currently living through a bizarre and deeply fragmented internet paradox.
Open any platform you use daily—be it social media, technical blogs, or news clients—and you will find pages increasingly flooded with AI-generated or AI-assisted content. At the same time, faced with this exponential explosion of information overload, human reading habits have undergone a fundamental shift. We no longer patiently click through links on traditional search engines; instead, we rely on a new generation of AI search engines like Perplexity and SearchGPT (now deeply integrated into ChatGPT) to browse, filter, and summarize the web for us.
This should have been a highly efficient intelligence loop: AI generates information → AI digests information → Humans extract value.
In reality, however, this loop has broken right down the middle. The pipelines of information are suffering from a severe case of “gridlock.”
1. The Fragmented Reality: Walled Gardens and a Fractured Loop
What blocks this loop is the legacy inherited from the Web2 era: the “walled garden.”
Traditional internet platforms—such as Reddit, X, WeChat Official Accounts, and mainstream media outlets—are facing unprecedented anxiety. They view the web scrapers and traffic from AI search engines as “data thieves,” arguing that these tools carry away the very assets they rely on for survival without paying a dime.
Consequently, an unprecedented “walled garden movement” and a fierce battle over interests are playing out globally. Major platforms are tightening their robots.txt rules and stacking up CAPTCHA challenges. The confrontation among tech giants has escalated into a full-blown war:
- Conflict and Litigation: Numerous traditional news publishers and platforms have publicly boycotted Perplexity, accusing it of bypassing anti-scraping protocols to “forcefully” grab content behind paywalls, even taking the matter to court.
- Acquisition and Alliance: OpenAI, backed by deep pockets, chose to pave the way with cash to protect SearchGPT. They have poured hundreds of millions of dollars into signing exclusive or high-stakes official data licensing agreements with media giants like Reddit, Condé Nast, and News Corp.
This current state reveals a highly inefficient dead end: content is exploding, users want to read it, but AI scrapers are left groping in the dark outside the walls. Independent developers or smaller AI tools that lack the funds to strike data deals are left with search results plagued by gaps and hallucinations. The original open spirit of the internet is actually regressing in the AI era due to disputes over value distribution.
2. The Next Paradigm: What is an “AI-Friendly” Platform?
Neither aggressive litigation nor massive exclusive buyouts can fundamentally solve the liquidity problem of the entire internet ecosystem. Since the trend of “AI producing content, AI consuming content” is irreversible, the underlying infrastructure of the internet is bound to face a massive reshuffle.
The logical next step is the rise of a new generation of “AI-Native” or “AI-Friendly” information platforms. Instead of treating AI scrapers as enemies, they will treat them as their core users. These platforms will feature three characteristics that completely subvert the tradition:
From “Anti-Scraping” to “Seamless API Integration”
Traditional websites are designed to please human eyes, packed with complex front-end code and user-verification puzzles explicitly meant to deter machine scraping. An AI-friendly platform, by contrast, is built from day one on a Machine-Readable architecture. The site will not only provide a graphical user interface for humans but will also natively feature a structured, semantic data stream interface (similar to an advanced version of JSON-LD or a dedicated RSS for agents). When an AI comes to scrape, it won’t need to go through complex webpage parsing; it can seamlessly read the content at lightning speed.
Diversity and “Multi-Modal Heterogeneity” of Content
These new platforms will completely break the constraints of traditional forums or blogs. The content hosted on them will be highly diverse and multidimensional. Text, code, dynamic charts, audio, video, and various intermediate data states generated by AI will be bundled together in formats perfectly optimized for cross-modal reasoning by large models. The platform will be more than just a “repository of words”; it will be a multi-modal sandbox natively built for agent throughput.
From “Capturing Attention” to “Agent Distribution”
The logic of a Web2 platform is to find every way possible to glue human eyeballs to the screen in order to sell ads. The logic of an AI-friendly platform is to find every way possible to “be retrieved and cited by AI.” Whichever platform provides higher quality and more transparent information to AI will be prioritized, digested, and ultimately presented to the end user by the agents of major models. The value of a platform will pivot entirely from “Page Views (PV)” to “API Citation Frequency.”
3. Commercial Reshaping: If AI Reads Everything, How Do Creators Survive?
A critical question arises: If everyone uses AI search to read summaries and no one actually clicks into websites anymore, how will platforms and content creators make money?
This is precisely where the core commercial opportunity for AI-friendly platforms lies. To break the deadlock, Perplexity has already begun experimenting with a “Publisher Revenue Share Program”—guaranteeing that whenever an AI answer cites a publisher’s content, a portion of the ad revenue is automatically shared with the original creator.
Within this new ecosystem, a more thorough value-circulation loop will be established:
- Micro-payments and Citation Revenue Sharing: When an AI tool retrieves high-quality content from a new platform and presents a summary to the user, every “citation” will trigger an underlying settlement protocol. AI companies will automatically pay tiny copyright or data service fees to the platform and the original author via micro-payments.
- The New “AI SEO” (Agent Optimization): Future businesses and creators won’t need to study how to please traditional search engine algorithms. Instead, they will focus on making their open data more understandable to AI. Whoever provides the most precise, authoritative, and heterogeneous data on an AI-friendly platform will capture the top spots on AI recommendation lists, thereby driving precise commercial conversions on the end-user side.
Conclusion: From Web for Humans to Web for Agents
The early internet thrived on the open nature of HTTP and RSS protocols, while the mid-era internet retreated into walled gardens driven by traffic monopolies. Today, AI is forcing the internet to fling its gates open once again.
We are accelerating away from a “Web for Humans” and moving toward a “Web for Agents”—a network where intelligent agents can freely navigate, propagate, and exchange value.
High walls cannot block a technological tsunami. Legacy platforms that refuse to open their doors to AI will eventually face a slow death as isolated information islands. Conversely, the first batch of truly open, AI-friendly platforms that embrace machines and accommodate diverse content will become the absolute foundation of the next-generation internet.
Leave a Reply