Need to let loose a primal scream without collecting footnotes first? Have a sneer percolating in your system but not enough time/energy to make a whole post about it? Go forth and be mid: Welcome to the Stubsack, your first port of call for learning fresh Awful youāll near-instantly regret.
Any awful.systems sub may be subsneered in this subthread, techtakes or no.
If your sneer seems higher quality than you thought, feel free to cutānāpaste it into its own post ā thereās no quota for posting and the bar really isnāt that high.
The post Xitter web has spawned soo many āesotericā right wing freaks, but thereās no appropriate sneer-space for them. Iām talking redscare-ish, reality challenged āculture criticsā who write about everything but understand nothing. Iām talking about reply-guys who make the same 6 tweets about the same 3 subjects. Theyāre inescapable at this point, yet I donāt see them mocked (as much as they should be)
Like, there was one dude a while back who insisted that women couldnāt be surgeons because they didnāt believe in the moon or in stars? I think each and every one of these guys is uniquely fucked up and if I canāt escape them, I would love to sneer at them.
(Credit and/or blame to David Gerard for starting this.)
yet again, you can bypass LLM āprompt securityā with a fanfiction attack
https://hiddenlayer.com/innovation-hub/novel-universal-bypass-for-all-major-llms/
not Pivoting cos (1) the fanfic attack is implicit in building an uncensored compressed text repo, then trying to filter output after the fact (2) itās an ad for them claiming they can protect against fanfic attacks, and I donāt believe them
I think unrelated to the attack above, but more about prompt hack security, so while back I heard people in tech mention that the solution to all these prompt hack attacks is have a secondary LLM look at the output of the first and prevent bad output that way. Which is another LLM under the trench coat (drink!), but also doesnāt feel like it would secure a thing, it would just require more complex nested prompthacks. I wonder if somebody is just going to eventually generalize how to nest various prompt hacks and just generate a āprompthack for a LLM protected by N layers of security LLMsā. Just found the āwell protect it with another AI layerā to sound a bit naive, and I was a bit disappointed in the people saying this, who used to be more genAI skeptical (but money).
Now Iām wondering if an infinite sequence of nested LLMs could achieve AGI. Probably not.
Days since last ānovelā prompt injection attack that I first saw on social media months and months ago: zero