A high -style war has just split about the future of the Internet. In one corner is Cloudflare, a giant of online infrastructure, which serves as a porter for a huge portion of Internet traffic. In the other there is perplexity, a darling of the AI world, a search engine threatening to take on Google’s domain.
The accusation is explosive: Cloudflare claims that perplexity is a bad actor, a rogue bottle that ignores the oldest rules of the Internet to secretly scrape data from websites that explicitly said it should stay away. The response of perplexity is equally fiery: it says that Cloudflare is either dangerously incompetent or engages in commercial, fundamentally misunderstanding how modern AI works.
The feud is the first major battle in conflict that will define the following era of the site: who reaches access to online information, and who decides the rules?
The charge: rogue -bot in disguise
For decades, the Internet has been operating on a “gentleman agreement” called robots.txt file. It is a simple text file that web site owners use to post digital “not enter” a sign for automated network crawls or “bots.” Well behavioral bots, like Google, respect this sign.
In a frightening Blog postCloudflare claims that perplexity ignores it. The company claims that when its stated bot, “perplexitybot”, is blocked, the AI search engine changes to stealth mode, using generic browser identities and rotating IP addresses to continue ramping and collection of data.
Cloudflare says it tested this creating completely new, private websites with strict rules “no bots allowed”. Even so, they found that “perplexity still gave detailed information about the correct content hosted on each of these restrictive domains.” Based on this “wrong crawling behavior”, Cloudflare announced that it has now listed perplexity as a controlled bot and actively blocks its undecided crawls.
The countering: “you don’t understand how AI works”
Perplexity Answer It was Swift, accusing Cloudflare of “almost everything bad about how modern AI helpers actually work.” The company argues that it is not a traditional “bot” and that cloudflare misappates old rules to new technology.
The core of their argument is the difference between a bot and a user. A traditional bot, like Google, systematically crawls billions of pages to build a mass index for later use. User agent, perplexity claims, acts on behalf of a real person in real time. When you ask perplexity, its AI -agent receives the necessary information from the network at that moment to answer you. It does not store data; It acts as your personal research assistant.
“This is fundamentally different from a traditional online creeping, in which creeping people systematically visit millions of pages to build massive databases, whether someone asked for that specific information or not,” perplexity wrote in a detailed response. “When companies like Cloudflare misbehave using AI helpers as malicious robots, they argue that any automatic tool serving users must be a suspicious position that would criminalize email clients and browsers.”
Then came Bombshell’s anti-charge. Perplexity claims cloudly “fundamentally miscribated 3-6m daily requests” from third-party cloud browser service to perplexity, calling it “a basic traffic analysis, which is especially embarrassing for a company whose core business understands and categorizes website traffic.” Perplexity suggests that this is either “clever advertising” or sign that Cloudflare is “dangerously misinformed about AI’s base.”
Users on social media have been shared. “Perplexity only uses a proxy to get something that is already on the public website, to answer a user’s question. Fitting it as some attack is absurd. The public website must be public,” defended Tech’s founder Andrej Radonjic. Another user was more critical: “Perplexity, pretending to be a search engine, pretending to be AI, yet neither.”
Perplexity only uses a proxy to get something that is already on the public website, to answer the user’s question.
Casting it as any attack is absurd. The public website should be public.
– Andrej (@0xdrej) 4 August 2025
Who owns the open website?
This public feud sets the central tension of the AI era. AI -Starts such as perplexity require access to the vast data ocean on the open website to work and compete with giants like Google and Openai. Without it, they cannot give real -time, accurate answers. But website owners are getting more and more concerned about their content scraped without consent or compensation to train and operate these new AI models.
Cloudflare, choosing to block the undecided crawls of perplexity, effectively appointed itself as the AI data police, making decisions on what constitutes a “legitimate” online traffic. Perplexity warns that this could lead to a “two-level internet”, where access depends not on user needs, but on whether their chosen AI tool was “blessed by infrastructure controllers.”
The rules of the Internet are rewritten in real time. The old gentleman’s agreement breaks, and the battle between the gatekeepers and the innovators has just begun. The result will determine not only the future of AI, but the future of the open website itself.
