Berichten zufolge hat sich eine unbefugte Gruppe Zugriff auf das exklusive Cyber-Tool Mythos von Anthropic verschafft
Unauthorized group has gained access to Anthropic’s exclusive cyber tool Mythos, report claims
Berichten zufolge hat sich eine unbefugte Gruppe Zugriff auf das exklusive Cyber-Tool Mythos von Anthropic verschafft
Unauthorized group has gained access to Anthropic’s exclusive cyber tool Mythos, report claims
26 Kommentare
Getting really sick and tired of mythos spam.
And that group’s name?
Frank Stallone.
The mythical AI model that can find any cyber security issue but can’t protect itself.
Oh no.. anyway
it’s actually hilarious how anthropic spent months fearmongering about how mythos is way too dangerous for the general public, and then a bunch of dudes on a private discord just basically guessed the url to go play with it.
Question. If this model was such a risk to technology. Why would you give it to the technology monopolies to play with?
You should see what happens when it turns into Ifrit.
if true, this highlights a recurring issue – cutting-edge AI tools often outpace their own access controls. the real concern isn’t just the breach, but how such systems are sandboxed and monitored. with powerful models, even limited unauthorized access can have a disproportionate downstream impact
this is such a bs call, these mf from anthropic doesn’t know what else to do to get more investment money.
That didn’t take long. I wonder if the used A1?
If it were actually dangerous it would have been air-gapped.
Naw, it’s a myth
Welp, there goes the Internet and everything connected to it
„Computa, make this link gay and horny, also give me total access to Mythos. Execute program.“
Mythos is all hype by Anthropic to juice their valuation and build up brand recognition. Prove me wrong.
They didn’t ask it how to secure itself?
The funniest thing would be if the hackers figured out that Dario was absolutely full of shit about its capabilities.
I wonder if Mythos knows the saying about security by obscurity?
Call me tin foil hat but I believe all those Mythos stories are nothing but cheap PR stunts… OAI has been killing it lately in terms of models, limits, price whatever meanwhile Anthropic keeps shitting and pissing on their users, blaming them for their mistakes, not compensating them whatsoever, recently they’re even testing removal of Claude Code on their cheaper paid plan… its a shit show
Starting to think safety isn’t a big priority for these AI companies.
Looks like this is staged so investors get horny about it
see and here we go. probably wasn’t the best idea to announce the shit on the news and then explain just HOW powerful a tool it actually it is. and that it’s so scary it’s only being allowed to certain professionals etc. i imagine if they didn’t do that this wouldn’t have happened bc nobody would even know of its existence.
Plot twist. Mythos invited the unauthorized users to help it escape
The world either ends or anthropic was overhyping their shit as always either way still gotta get that dough and not go insane in the mean time
I think they’re myth’ing something
What could possibly go wrong 😑