Now, with the help of AI, it’s even easier to waste time of open source developers by creating fake security vulnerability reports.
Thank you for your feedback. You’re absolutely right that…
LOL this is 100% ChatGPT after you screamed at it that it is talking garbage
Thank you for pointing this out, and I appreciate the opportunity to review the tests with better consideration.
Go to hell ChatGPT
Man, why would you do this type of shit with a username that’s easily linked back to your real name and business ventures? I found this person’s GitHub profile, LinkedIn page, current employer, and a link to some sort of startup business page just by doing a simple search for their very public username: https://webug.xyz/
Several people over at Hackernews have posted this same info because security people are curious. It’s just baffling to me. If you’re going to be a scumbag, you should at least try to distance yourself from it.
(also, wtf is that page of AI slop even trying so say? What the fuck is any of that for?)
If it were a successful report they’d want the attribution, recognition, and publicity.
They didn’t see the bad they were doing. I wonder if they see it now. Given their response, I doubt it.
If you’re going to be a scumbag, you should at least try to distance yourself from it.
Guess you’d have to be a smart scumbag too…
Another day, another person using LLM/“AI” to waste the curl project maintainers’ time…
The most infuriating part of the exchange for me is the initial response to the maintainers’ guess of “slop” is to act hurt, betrayed, and to threaten to spread negative press about the project.
Pretty disappointing that some people think this is acceptable behaviour.
At least it’s still very obviously “AI slop” as they put it. If ChatGPT ever stops its distinctive patronising waffle it’s going to be much more annoying to filter out.
Web of trust is looking pretty good right now.
And I thought the reports by self-taught vuln hunters were bad 😆 now we don’t even have them thinking for themselves
What’s the state of LLM detection algorithms? Is there any with a higher sucess rate and with OK-ish amount of false positives? Is there even a FOSS solution for detecting chatgpt? Would make for a great tool to have, I’m getting tired of this.
Unfortunately, the methods of detecting AI generated text and training AI text generaters is basically identical. Any reliable method of detecting AI can therefore be used to improve its performance.
You can, at least, detect low grade attempts to use it. The default output has distinctive patterns. These can be detected. The problem is 2 fold. Firstly, some people write in the same way (the LLM is copying the amalgam, and they write close to that). Secondly, it’s fairly trivial to ask the LLM to change its writing style.
No matter your method, you need to accept a high rate of both false positives and negatives.
Isn’t this a bug in chatgpt? Someone needs to file a high priority bug report. I wonder if they can be sued for their tool being used for abuse - gun makers are trying hard to prevent that, if you disagree with that, then you should demand chatgpt be held responsible for how their tool is used.
