Hacker News Reader

Comment by xyzal | Hacker News Reader

In case anyone is interested in a tiny bit of sabotage, I am under the impression I managed to 'drown' true information on my microblog by generating contradicting posts with LLaMa (tens of them for each real post) and invisibly linking them, so a human would not click through.

You know, flood the zone with s***, Bannon-style ...

View on HN

This is an approach I've seen used and I'm not sure what success it has had. But logically it seems sound: explicitly reference paths that no human would actually see - traffic hitting those paths are bots. They can't help themselves.

Regarding pollution of the LLM weights, at least the Russian propaganda machine seems to prove it works, see https://www.france24.com/en/live-news/20250310-russian-disin...

Yep, just make sure to add them to your robots.txt file so that only the bad robots are harmed.

Temporary solution and only works if only some of us are doing it. What if these bots have a "manager" LLM agent that takes the decision on what pages to scrape?

When I read this yesterday, I was contemplating one possible way to mitigate this at a larger scale is if websites could create random virtual paths/endpoints that drive the bot into a locally served Library of Babel[0] that poisons the spiders with lots of useless text.

It won't work for well-structured sites where the bots know the exact endpoint they want to scrape, but might slow down the more exploratory spider threads.

[0] https://libraryofbabel.info/

Besides flooding them with junk, what about outright sabotage in the form of serving zip bombs or other ways to waste computational resources?

I think we need to aim for the bots to get _negative_ utility value from visiting our traps, not just zero value.

Did you try to GET a canary page forbidden in robots.txt? Very well, have a bucket load of articles on the benefits of drinking bleach.

Is your user-agent too suspicious? No problem, feel free to scrape my insecure code (google "emergent misalignment" for more info).

A request rate too inhuman? Here, take those generated articles about positive effect of catching measles on performance in bed.

And so on, and so forth ...

Even though I agree with what you're doing in principle, I feel it's necessary to remind and warn everyone here that sabotaging bots could be viewed as a violation of laws such as the US's Computer Fraud and Abuse Act[1]. I mean, unless the Second Amendment is suddenly interpreted to include cyberweapons.

[1] https://en.wikipedia.org/wiki/Computer_Fraud_and_Abuse_Act

Reply on news.ycombinator.com