Live data from Hacker News

Perplexity is using stealth, undeclared crawlers to evade no-crawl directives

blog.cloudflare.com

161–170 of 799 posts

Re: Perplexity is using stealth, undeclared crawlers to evade no-crawl directives

#161

"Stealth" crawlers are always going to win the game. There are ways to build scrapers using browser automation tools [0,1] that makes detection virtually impossible. You can still captcha, but the person building the automation tools can add human-in-the-loop workflows to process these during normal business hours (i.e., when a call center is staffed). I've seen some raster-level scraping techniques used in game dev…

> "Stealth" crawlers are always going to win the game.

no, because we'll end up with remote attestation needed to access any site of value

Re: Perplexity is using stealth, undeclared crawlers to evade no-crawl directives

#163
post #80

Earlier quoted context omitted.

I think it's an issue of scale. The next step in your progression here might be: If / when people have personal research bots that go and look for answers across a number of sites, requesting many pages much faster than humans do - what's the tipping point? Is personal web crawling ok? What if it gets a bit smarter and tried to anticipate what you'll ask and does a bunch of crawling to gather information regularly to…

Doesn't o3 sort of already do this? Whenever I ask it something, it makes it look like it simultaneously opens 3-8 pages (something a human can't do). Seems like a reasonable stance would be something like "Following the no crawl directive is especially necessary when navigating websites faster than humans can." > What if it gets a bit smarter and tried to anticipate what you'll ask and does a bunch of crawling to ga…

You don't middle click a bunch of links when doing research? Of all the things to point to I wouldn't have thought "opens a bunch of tabs" to be one of the differentiating behaviors between browsing with Firefox and browsing with an LLM.

Re: Perplexity is using stealth, undeclared crawlers to evade no-crawl directives

#164

Crawling and scraping is legal. If your web server serves the content without authentication, it's legal to receive it, even if it's an automated process. If you want to gatekeep your content, use authentication. Robots.txt is not a technical solution, it's a social nicety. Cloudflare and their ilk represent an abuse of internet protocols and mechanism of centralized control. On the technical side, we could use CRC m…

I think Cloudfare is setting themselves up to get sued.

(IANAL) tortious interference

Re: Perplexity is using stealth, undeclared crawlers to evade no-crawl directives

#165

Crawling and scraping is legal. If your web server serves the content without authentication, it's legal to receive it, even if it's an automated process. If you want to gatekeep your content, use authentication. Robots.txt is not a technical solution, it's a social nicety. Cloudflare and their ilk represent an abuse of internet protocols and mechanism of centralized control. On the technical side, we could use CRC m…

Would highly recommend listening to the latest Hard Fork podcast with Matthew Prince (CEO, Cloudflare): https://www.nytimes.com/2025/08/01/podcasts/hardfork-age-res...

I was skeptical about their gatekeeping efforts at first, but came away with a better appreciation for the problem and their first pass at a solution.

Re: Perplexity is using stealth, undeclared crawlers to evade no-crawl directives

#166
post #139

Earlier quoted context omitted.

It's all about scale. The impact of your personal shopper is insignificant unless you manage to scale it up into a business where everyone has a personal shopper by default.

Well then. Seems like you would be a fool to not allow personal shoppers then. The point is the web is changing, and people use a different type of browser now. Ans that browser happens to be LLMs. Anybody complaining about the new browser has just not got it yet, or has and is trying to keep things the old way because they don’t know how or won’t change with the times. We have seen it before, Kodak, blockbuster, wha…

> Anybody complaining about the new browser has just not got it yet, or has and is trying to keep things the old way because they don’t know how or won’t change with the times. We have seen it before, Kodak, blockbuster, whatever.

You say this as though all LLM/otherwise automated traffic is for the purposes of fulfilling a request made by a user 100% of the time which is just flatly on-its-face untrue.

Companies make vast amounts of requests for indexing purposes. That could be to facilitate user requests someday, perhaps, but it is not today and not why it's happening. And worse still, LLMs introduce a new third option: that it's not for indexing or for later linking but is instead either for training the language model itself, or for the model to ingest and regurgitate later on with no attribution, with the added fun that it might just make some shit up about whatever you said and be wrong. And as the person buying the web hosting, all of that is subsidized by me.

"The web is changing" does not mean every website must follow suit. Since I built my blog about 2 internet eternities ago, I have seen fad tech come and fad tech go. My blog remains more or less exactly what it was 2 decades ago, with more content and a better stylesheet. I have requested in my robots.txt that my content not be used for LLM training, and I fully expect that to be ignored because tech bros don't respect anyone, even fellow tech bros, when it means they have to change their behavior.

Re: Perplexity is using stealth, undeclared crawlers to evade no-crawl directives

#167
post #139

Earlier quoted context omitted.

It's all about scale. The impact of your personal shopper is insignificant unless you manage to scale it up into a business where everyone has a personal shopper by default.

Well then. Seems like you would be a fool to not allow personal shoppers then. The point is the web is changing, and people use a different type of browser now. Ans that browser happens to be LLMs. Anybody complaining about the new browser has just not got it yet, or has and is trying to keep things the old way because they don’t know how or won’t change with the times. We have seen it before, Kodak, blockbuster, wha…

Do not conflate your own experience with everyone else's.

Re: Perplexity is using stealth, undeclared crawlers to evade no-crawl directives

#168

I find this problem quite difficult to solve: 1. If I as a human request a website, then I should be shown the content. Everyone agrees. 2. If I as the human request the software on my computer to modify the content before displaying it, for example by installing an ad-blocker into my user agent, then that's my choice and the website should not be notified about it. Most users agree, some websites try to nag you into…

Nothing wrong if they fetch on your behalf. The problem is when they endlessly crawl along with every other ai company doing the same.

Re: Perplexity is using stealth, undeclared crawlers to evade no-crawl directives

#169
post #30
post #22

Earlier quoted context omitted.

You're free to deny access to your site arbitrarily, including for lack of compensation.

Like for people or are using a ad block or for a crawler downloading your content so it can be used by an AI response?

Arbitrarily, as in for any reason. It's your site, you decide what constraints an incoming request must meet for it to get a response containing the content of your site.

Re: Perplexity is using stealth, undeclared crawlers to evade no-crawl directives

#170

I find this problem quite difficult to solve: 1. If I as a human request a website, then I should be shown the content. Everyone agrees. 2. If I as the human request the software on my computer to modify the content before displaying it, for example by installing an ad-blocker into my user agent, then that's my choice and the website should not be notified about it. Most users agree, some websites try to nag you into…

I don't think people have a problem with an LLM issuing GET website.com and then summarising that, each and every time it uses that information (or atleast, save a citation to it and refer to that citation). Except ad ecosystem, ignoring them for now, please refer to last paragraph. The problem is with the LLM then training on that data _once_ and then storing it forever and regurgitating it N times in the future wit…

> LLMs should also use citations.

Mojeek LLM (https://www.mojeek.com) uses citations.

Post reply on HN