Designing impossible mazes for AI to get lost in as a matter of civil defense? What a fucking world we live in…
post
Man do I hate the age we live in now, where people are actively having to declare war on scrapers like this because AI bros have literally zero respect towards the rest of the internet.
I just hope all that endless maze scraping cost some asshat a decent chunk of money.
Right! This isn't "AI going rogue", it's software agents made and employed by perfectly human psychopaths, doing what they were made to do.
I just hope all that endless maze scraping cost some asshat a decent chunk of money.
Unfortunately, it probably cost the maze's owner nearly as much, due to all the traffic.
What if instead of trying to cost them directly with crawling, you made what they crawl worthless? Could somewhat ironically use an LLM to generate gibberish output for them to feed on. Or a bash script can do it too.
Is ironic the word for OP being a repost bot? Or is it just expected at this point?
You know, I wonder if the AI agents are willing to open random zip files they find on websites, and I'm further curious if they're smart enough to avoid a zip bomb.
Can follow JWZ for practical experience on this kind of thing - apparently it is getting so bad that even opening a connection to deliver the zip bombs is getting infeasible now.
You can serve zip bombs as web pages since most crawlers support gzip encoding.
And yes it's very effective ;)
So what you're implying is this should be implemented in every website in both follow/not follow just to waste their money?
[Everyone liked this.]
I wonder if it's possible to trap bots like this and use them to mine crypto.
No, you didn’t.
Yeah I have OP listed as "repost bot" so if I see something cool I can go check somewhere else for the info.
This was a fun story. I masturbated. I did not finish, nor did I enjoy myself. At the same time, I have procured a thought sapling about slime molds, specifically how a specialized biological system could outperform a binary computer. This triggers a thought about quantum computing being better at logistics but shit at counting, for example. Information density directly correlates with the degree of topological entanglement, which means the möbiation of entanglement is itself the mechanism to not just encode additional density of information, but this information encoded can be parsed differently by different systems, to thus allow a multivariable set of instructions to exist in the same space. Thank you, I will take my Nobel Prize as a suppository.
Big if true
Is it possible to have the bots recursively give themselves instructions and birth child instances of themselves that do the same thing? Because if it's ai bots, that could absolutely be maliciously exploited.
Can anyone explain what is happening behind the scenes with the "room creation"?? That sounds intriguing. I understand OP is a bot but, is anything described here an actual way of trapping crawlers? How are the crawlers fooled by any of this?
He placed an invisible link on his website. Only visible by reading the page content (html). Then the bots followed the link. This link lead to a page with 5 links (door) four links are dead and one leads to another page (level) with again 5 links.
I guess the links and pages are generated on the go. So the bots and crawlers may follow the links forever of they like.
I wonder if pages have to be generated individually to appear legit to the crawlers, or if the same contents can be served as different links over and over... if I knew more about how this stuff works I'd try it myself hah.
Most crawlers don't care - they're purely downloading. Most garbage will also be filtered out before training, and generating plausible garbage is expensive - not really worth the trouble. What you can do with the garbage, is that you can place unique identifiers in the links you serve, so when a crawler that you can't recognize otherwise comes visiting a poisoned URL, you can firewall it off. This works because the crawlers use a queue: they get a list of URLs to visit, they download them, and later on, the URLs from those get extracted, and crawlers will visit those URLs.
So if you manage to trap a bot, and overwhelm its queue with poisoned URLs, you can also catch bots that you wouldn't be able to easily identify otherwise. And that is why serving an endless maze of garbage is so very useful. Because it lets you catch stuff like headless chromes operating through residential proxies: because those too, follow a queue.
As some other users have already mentioned it, have a look at this project anyway: https://iocaine.madhouse-project.org/
It gives much more bloated information for a crawler than I expected. But sounds quite fun to host.
Apparently some bot programmers are assholes who ignore rel="nofollow".
"You" did no such thing.
Account blocked. Bot, or not.
It's ironic that you advertise yourself as having "time to lose" but you don't have "time to replace 1st person pronouns with more accurate 3rd person ones."
Throw in a few of them torture chambers and make it a proper dungeon >:)

top 50 comments