LLM scraper bots are overloading acme.com's HTTPS server
acme.com43 points by mjyut 4 hours ago
43 points by mjyut 4 hours ago
> The LLM companies are not picking on me in particular, they are pounding every site on the net.
Why is not this a criminal offense? They are hurting business for profit (or for higher valuation as they probably have no profit at all).
Why are corporations allowed to do with impunity what could land even a teenager years in prison? Is there no rule of law anymore?
The five-year and ten-year penalties kick in only when the government can show the offense caused at least $5,000 in losses across all victims during a one-year period. https://legalclarity.org/what-are-the-punishments-for-a-ddos...
> Why are corporations allowed to do with impunity what could land even a teenager years in prison? Is there no rule of law anymore?
Those laws are intended to protect corporations. If corporations are the ones doing the scraping, it doesn't make sense for the same laws to affect them.
Normative vs prerogative state [1]. See US v. Swartz compared to Meta use of LibGen for Llama
So, I knew Aaron and I definitely would not presume to predict what he would have thought, but I’d point out there is a sizeable state space where he should never have been prosecuted, and scraping by others including large commercial companies should not prosecutable on the same grounds.
I repeat what Aaron’s friends and lawyers said at the time: we were going to fight that case, and we were going to win.
Because might makes right and any entity with the power to legally put up a fight is in on the game (or wants to be)
I have added a DB replica server just to keep my website from succumbing to AI bot traffic.
Because they have more money.
I've had to deploy a combination of Cloudflare's bot protection and anubis on over 200 domains across 8 different hosting environments in the last 2 months. I have small business clients that couldn't access their sales and support platforms because their websites that normally see tens of thousands of unique sessions per day are suddenly seeing over a million in an hour.
Anthropic and OpenAI were responsible for over 70% of that traffic.
It's a bit more like a physical business with a "public welcome" policy like a coffee shop going viral and then having tens of thousands of people walking in and taking pictures but not buying coffee. It's disruptive, but not illegal.
Acme.com is welcome to require authentication for all pages but their home page, which would quickly cause the traffic to drop. They don't want to do this - like the coffee shop, they want to be open to public, and for good reasons.
Sometimes the use profile changes dramatically in a short time. 15 years ago, Netflix created the video streaming market and shared bandwidth capacity that had been excessive before wasn't enough. 15 years before that, Google did the same thing when they created search and started driving tremendous traffic to text based websites which had spread through word of mouth before.
Turns out the micro transaction people probably had the right idea.
> Is there no rule of law anymore?
Have you not been paying attention to the news for the past few years?
No, there isn't. If there were, Trump would be in prison, not the Oval Office. And he and the Republican Party have deliberately fostered this environment of corruption and rule-by-wealth so that they can gain more power and even more wealth.
And now they are also backing the AI zealots, and techbros more generally, to ensure that they can do whatever the hell they want, damn the consequences to the rest of the world.
We've already established that computer crime and IP laws apply to normies and not tech companies
Is what an offence lol? Bot scraper traffic?
How do you think search engines work?
Search engines appear to care more about being good "Netizens". It's not like GoogleBot never crashed a site, but it's rare. Search engine bots check if they need to back off for a bit, they check etags, notices if page changes infrequently and slow down their crawler frequency.
If you train an LLM, it's not like you keep a copy of every page around, so there's no point to check if you need to re-scrape a page, you do, because you store nothing.
Personally I think people would be pretty indifferent to the new generation of scrapers, AI or other types, if they at least behaved and slowed down if they notice a site struggling. If they had the slightest bit of respect for others on the web, this wouldn't be an issue.
They work because they offer ways to opt out, they honor crawl delay, setting ideal scraping times, IndexNow, etc.
And they give you real, valuable traffic in return.
Because the law deals with intent. The intent for a 12 year old skiddie with a ddos box is to harm someone else's internet. the intent of big scrapers is to collect data. if you want to make the latter illegal then vote for that instead of loading it with the normative baggage of the former.
It's the same problem as why Occupy Wallstreet fell apart: bunch of losers who don't understand the system screech about the system. because they don't understand it, they can't offer any meaningful dialogue about how to fix it beyond screeching.
I suspect part of the issue is that people are still using things like `acme.com` and `demo.com` as an example domain in their documentation and tests instead of relying on `example.com` which is reserved exactly for this purpose [0]
A small part. On my server AI bots outnumber real visitors 300 to one.
I don't mean that users are following the links to `acme.com` and `demo.com` type domains in documentation; I mean that bots are likely finding and following many links to them because of their widespread use in documentation.
If you search for `site:github.com "acme.com"` in Google, you'll find numerous instances of the domain being used in contrived links in documentation as an example of how URLs might be structured on an arbitrary domain and also in issues to demonstrate a fully qualified URL without giving away the actual domain people were using.
This means that numerous links are pointing to non-existent paths on `acme.com` because of the nature of how people are using them in documentation and examples.
That is very possible.
But it is not necessary to see the results that are being described.
If sites like my tiny little browser game, with roughly 120 weekly unique users, are getting absolutely hammered by the scraper-bots (it was, last year, until I put the Wiki behind a login wall; now I still get a significant amount of bot traffic, it's just no longer enough to actually crash the game), then sites that people actually know and consider important like acme.com are very likely to be getting massive deluges of traffic purely from first-order hits.
That such an absolutely ludicrous thing to hear in a "wtf are these people doing" type of way. I can't imagine a non-social media site would be generating enough traffic to the level that these bots need to be essentially doing continuous scraping. It's just gross to me to be okay with that level of unsophisticated effort that they just do the same thing over and over with zero gain.
How are you measuring this? Does your solution rely on user agent or device fingerprinting? Curious to know what tools are available today and how accurate they are.
Where from? And quite frankly why? There are existing training data sets that are large enough for smaller models. Larger models have been focusing on data quality more than quantity. There's limited utility to further indiscriminate widespread scraping,
Tell that to the idiots doing the scraping.
Small site operators like us know very well that the utility they can get by scraping us is marginal at best. Based on their patterns of behavior, though, my best guess is that they've simply configured their bots to scrape absolutely everything, all the time, forever, as aggressively as possible, and treat any attempt to indicate "hey, this data isn't useful to you" as an adversarial signal that the site operator is trying to hide things from them that are their God-given right.
i had to block all traffic, except that of my country. As i offer a service that is exclusive to my country it worked like a charm.
For those who have deployed Cloudflare in front, what are pros and cons? How's the user experience? Do they offer free bot protection?
I opted for Bunny Shield exactly to combat bots, in particular ones that spoof User Agents and rotate millions of IPs. It works great, detecting the vast majority of bots and challenging them. Much more user friendly than Cloudflare too, which typically resorts to challenging everyone (not that CF was ever an option due to various concerns).
Bot traffic is crazy even for smaller sites, but still manageable. I was getting 2,000 visitors a day on my infrequently updated website, but after I blocked all the bots via Cloudflare it went back to the normal double digit visitor count.
I have 6M pages across 8 domains. I have 10 unique IP residential bots per second working hard to scrape every single page.
and how often are those 6M pages changing? how often are those bots finding anything new? why are the bot makers not noticing no difference and just slowing the request down for essentially stale content to them
In March 2025, Drew DeVault wrote a blog post called "Please stop externalizing your costs directly into my face"[1]. I think that is a pretty good guess as to why these bots do not care about frequency of changes, it costs to much.
Every run is basically a fresh run, no state stored, every page is just feed into the machine a new. At least that's my theory.
The AI companies need a full copy of your page, every time they retrain a model. Now they could store that in their own datacenters, but that's a full copy of the internet, in a market where storage costs are already pretty high. So instead, they just externalize the storage cost. If you run a website, a public Gitlab instance, Forgejo, a wiki, a forum, whatever, you basically functions as free offsite storage for the AI companies.
1) https://drewdevault.com/2025/03/17/2025-03-17-Stop-externali...
On the platform at my work they scrape the same page multiple times, over and over. They do not care to cache anything. And it’s ridiculous to account for because for example for our properties, everything is news-based so warming the cache was as simple as loading the first X articles to get them into cache. But with AI that is not viable because they scrape as much as possible, articles from 2018, 2017. Management doesn’t want to block them though. It’s just suffering through the endless barrage. I was able to do a lot for this like heavier caching even with pgpool but it’s so crazy that this small subset of bots effectively accounts for like 60%+ of our spend.
> Nearly all of them were for non-existent pages.
Do any webservers have a feature where they keep a list in memory of files/paths that exist?
Also why are most requests for non existent pages?
Because they are hunting for vulnerable devices and the requests' existence are unique to an application. Like a VoIP appliance for example.
They usually request something deep like /foo/bar/login.html as part of their reconnaissance.
I'm up to 4 pages of filter rules after the massive IP blacklist.
These assholes are also scanning every address on the IPv4 internet and hoovering up the content.
To answer your first question: No, that's the OS's job. But some clever rules could be setup for filtering invalid requests depending on your web server.
The only real solution is to put Anubis in front. For me, I just use Cloudflare in front and that suffices. But it's only a few thousand per hour by default. My homeserver can handle that quite well on its own.
Series of Chinese LLM scrapers kept PortableApps.com running slow and occasionally unresponsive for 2 weeks.
There are plenty of local LLMs out there run by humans that play nice. It's not the LLMs that are the problem. It's the corporations. That's the commonality. Human people aren't doing this. These corporate legal persons are a much more dangerous and capable form of non-human intelligence with non-human motives than LLMs (which are not doing the scraping or even calling the tools which are sending the HTTP requests). And they have lobbied their way to legal immunity to most of their crimes.
> Human people aren't doing this
Who do you think writes these scrapers? Well, I mean aside from the vibe coded ones.
> Someone really ought to do something about it.
What is bro proposing here?