Prompt injection risk/avoid

Top Creators

Public Insight Cards

These are deterministic, source-backed cards from the offline export. They are not live AI answers.

Related Source Records

Evidence Passages

I just heard a really interesting idea on how you might be able to trick ChatGPT into mentioning your website, and so I just wanted to share it with you. It's actually just a simple update to your robots.txt file. If you don't know, this is a file that lives on every website, and it tells bots like Google Bot or ChatGPT bot which pages of their website they're allowed to crawl. And all the big robots really do follow these instructions. So if you block ChatGPT or you block Google from crawling your website, you...

Problem is that as I understand it, these large language models aren't actually reading your robots.txt file. Instead, it's calling a web crawler tool. That tool is what actually reads the robots.txt file and decides if it can crawl the website. The only thing ChatGPT is seeing is what that web crawler returns after crawling the website. All that being said, I was very impressed with the video from Nate B. Jones, so I'll link to that here. And the punchline is that none of us really know what's gonna work in the...

Business owners have been tricking Chat b t into recommending their business using some pretty funny tactics. So I just thought we'd talk about one of them, prompt injecting. So this tactic stems from how large language models like ChatGPT sometimes retrieve information. If you ask for anything that requires them to do a search, they'll often do a dozen or more searches. They'll look through any web pages that seem relevant to your request and then they will return passages from those web pages which will be add...

awl your website because again, it's common for the tools to do dozens of searches and often look through hundreds of search results. You might say, sure, that's working right now, but surely the large language models will be smart enough to ignore that in the near future. But that's not so clear. In fact, I have one friend that I trust for this kind of stuff. His name is Steve. Hi, Steve. If I understand his position correctly, he's convinced that large language models will always be susceptible to prompt injec...

some kind of manual process that reviews the offending pages or websites and just removes them from the index. So if Steve's right and this problem really is unsolvable, my guess is that these websites using prompt injection are gonna be removed from the index that the L L m's use to retrieve search results. But that's what's crazy about the time we're living in. We really have no idea. It is truly the Wild West right now. It's a little bit scary but really exciting.