Quincy ⁂

"LLM-generated passwords"

"LLM-generated passwords fundamentally weak, experts say" and it's really a headline

I can't even.

I. can't. even.

I can't even begin to apprehend the #brainrot that has been pushed on humanity by "AI" terrorists.

Every time I think I get it, something like this comes across.

Beam me up.

Feb 19 at 5:20amWeb
Show threadQuincy ⁂Feb 19
"Motor oil fundamentally not safe to drink, experts day"
Show threadDanielFeb 19
@quincy Ah, what have experts ever done for us?
Show threadQuincy ⁂Feb 19
@anotherdaniel Sometimes they have to state the obvious :D
Show threadQuincy ⁂Feb 19
@anotherdaniel and still not be believed (cf. coronavirus)
Show threadDanielFeb 19

@quincy In the US, they have managed to create a population that is so lost, no one I speak with can even imagine how to turn that around again.
And thanks to most of the rest of the world lapping up every tidbit of moronic idiocy coming out of that broken culture, we are following hot on their heels.

And all you can do is watch, and listen to your brain scream.

Show threadQuincy ⁂Feb 19
@anotherdaniel indeed :(
Show threadQuincy ⁂Feb 19

How on earth ...

but education is slow, whereas lies and hypes are quick.

and now even quicker, due to "#GenAI".

Show threadQuincy ⁂Feb 19

"GenAI" needs to be abolished entirely, the sooner the better.

And those pushing it, especially in the education "sector" are culpable of crimes against humanity, IMO.

Crimes yet to be defined, maybe

But crimes all right.

Show threadQuincy ⁂Feb 19

Education needs to fortify people against this kind of deceit.

Instead, it's being commandeered by the enemy.

All in the name of "technological progress".

Madness!

Show threadQuincy ⁂Feb 19
(using the neoliberal phrase "the education sector" made me cringe the instant I wrote it)
Show threadC.Suthorn Feb 19
@[email protected] ich als großes Sprachmodell sage "12345678" ist ein besonders verbreitetes Passwort. Ich kann Dir gerne weitere beliebte Passwörter generieren, sag mir einfach die website und Deinen loginnamen.
Show threadQuincy ⁂Feb 19
@Life_is Au ja! Und bitte sag mir, ob dieses Passwort nicht einfach überall benutzen kann, wäre da nicht viel einfacher?
Show threadRyan Castellucci (they/them) Feb 19
@quincy if anything they might be useful for cracking passwords
Show threadMike JohnstonFeb 19
@quincy may as well just grab a random one from rockyou.txt 🤣
Show threadvampirdaddyFeb 19
@quincy Passwords are negible anyway as ClaudeVot/ClawdBot will use and leak that regardless.
Show threadMike McCaffreyFeb 19
@quincy Funny how the random generation machines can't even get random generation right.
Show threadRejZoRFeb 19
@quincy Why do we need LLM to generate some random shit? It already does that by default.
Show threadQuincy ⁂Feb 19

@rejzor I have no idea.

Or rather, I do.

There's a chatbox that claims to replace all specialized tools, so people use it ...

Show threadFrank Bennett 🇯🇵Feb 19
@quincy Kind of shocked people need to have this one pointed out to them. It's like saying software runs on computers.
Show threadTattieFeb 19
@quincy this is like "here is a random number I generated, feel free to use it for all your random-number needs: 11"
Show threadlil5Feb 19

@quincy

This is python’s math/rand vs crypto/rand all over again, except this time nobody reads the fucking documentation!

Show threadSpitefully OptimisticFeb 19
@quincy So you tasked the model that is supposed to come up with the likeliest string of characters to a problem with giving you a random password? Well, I got news for ya... :D
Show threadlizbianFeb 19
@quincy Wait, wait. The system, that is meant to predict what comes next after seeing a text, can only generate predictable passwords? Who would have thought!
Show threadadorferFeb 19
@quincy "Always check that your passwords have 0 google hits" <g>
Show threadargonautFeb 19
@quincy Uhm... who has made this a headline? Would like to read the source.
Show threadQuincy ⁂Feb 19

RE: https://code4lib.social/@dltj/116095562187115384

@argonaut It's from TheRegister: https://chaos.social/@[email protected]cial/116095562246197994

Show threadDFeb 19
@quincy It's hard to believe that the statistical plausibility machine might not be a good source of entropy.
Show threadMarjorieRFeb 19
@quincy no doubt the LLM had been trained on the 'I've been Pwnd' dataset.
Show threadQuincy ⁂Feb 19
@marjolica very likely!
Show threadKevin Karhan Feb 19

@quincy seriously, I thought we had that re: password security & cryptogeaphy already done...

At leadt I did…

Show threadjan Ki | 奇  Feb 19
@quincy
I have questions I don't wanna know the answers to
Show threadAshwin DixitFeb 19

@quincy

Famous last passwords.

[stolen]

Show threadHugo Slabbert ⚠️Feb 19
@quincy
"We have a task where one of the primary measures of success is entropy. Let's give it to the entropy flattening machine."
Show threadSophie 🐈 🇨🇦Feb 19
@quincy umm actually sweaty the more you boil the ocean the more secure your password is
Show threadTorf und SchneeFeb 19
@quincy "LLM-generated passwords" Wait...WWAT???
Show threadzl2todFeb 19

@quincy

Every externally generated password comes compromised.

Show thread~nFeb 20
@quincy lel... theregister: "Generative AI tools are surprisingly poor at suggesting strong passwords, experts say."
yes. "surprisingly" - wtaf.
Show threadQuincy ⁂Feb 20
@nblr "We tasted 10 brands of motor oil and they all tasted surprisingly bad."