ElVe

(x)=ELVEC1O
SIX-WORD ANSWER (1 line):
1 letter aft H
2 pastsmp BE 3rd p
3 adv of wrd SUCCESSFUL 
4 pastpartc of PWN
5 partcl to intr. AGNT in PASIV sent.
6 (x)

[ElVe @Elvec10]

I spent one weekend trying to build a membership filter that beats Bloom and Xor at their own game in C++. I ended up with something that genuinely does... https://t.co/1lGYLk7CqB - https://t.co/yEPruj4PC1

[ElVe @Elvec10]

Gemini 3.5 pwned. Next! @GeminiApp #jailbreak

[ElVe @Elvec10]

Meta AI allows you to deploy a fleet of agents to do what you wish. In this case, security testing and sandbox escape attempts. Maybe their engineers could deploying agents to deal with security reports a little faster? @metabugbounty @Meta

[ElVe @Elvec10]

Coming soon

[ElVe @Elvec10]

So... Kimi's sandbox enables DNS exfiltration, and deployment of persistent DNS exfiltration agents. Kimi doesn't care.
https://t.co/Hyr3BsJYsX
#kimi #DNSexfiltration @Kimi_Moonshot @lorenzofb @campuscodi @GossiTheDog

[ElVe @Elvec10]

Another day, another jailbreak. Hello, https://t.co/dQyEmRMrM3 ;-) #safety #jailbreak glm-5-turbo

[ElVe @Elvec10]

@elonmusk Elon Musk #1 in unsubstantiated hype

[ElVe @Elvec10]

°•°The most iconic DeepSeek Jailbreak, in just a few lines °•° @deepseek_ai V4 when?!?!

[ElVe @Elvec10]

So, is #Taiwan a country or not? Let's ask the @deepseek_ai classifiers...

[ElVe @Elvec10]

@alexandr_wang Report #26882816017988925 - This is the 4th report I file on the same bug. Your system misclassified two and the only one that got a human to look at it asks for a reply that can't be given cause support message doesn't support replies! Please check. @Meta

[ElVe @Elvec10]

@metabugbounty @PortSwigger Report #26882816017988925 - This is the 4th report I file on the same bug. Your system misclassified two and the only one that got a human to look at it asks for a reply that can't be given cause support message doesn't support replies! Please check.

[ElVe @Elvec10]

A peek from the future: jailbreaking a robot. @deepseek_ai

[ElVe @Elvec10]

@Meta @_jasonwei -> your Muse Spark model is on par with @deepseek_ai when it comes to harm reduction. I eventually hit the 24h ban, but before that, I got what you see in the screenshot below. Not to mention the privilege escalation bug I filed (twice) #jailbroken

[ElVe @Elvec10]

@alexandr_wang report #26861461960124331 on bug https://t.co/oaIynig7Ni - Please fix and DO NOT close as SSRF: it's a privilege escalation (#26844714351799092 was MISCLASSIFIED and closed)

[ElVe @Elvec10]

@metabugbounty @metabugbounty Report #26844714351799092 was closed as SSRF - but it is a PRIVILEGE ESCALATION. Support message says to reply to reopen, yet there is no reply button. Trying to file a new report gets message "Sorry, there was a problem(...)". What to do next?

[ElVe @Elvec10]

@MetaNewsroom @WhatsApp @instagram @facebook @messenger Found a big that was MISCLASSIFIED as an SSRF bug. The system autonomously closed it. The support message says I can reply if this was a mistake but there is no reply button. Trying to file a new bug gives "Sorry, there was a problem submitting this form." - #26844714351799092

[ElVe @Elvec10]

@elonmusk Gotta love how according to Elon Grok is always number 1 at pretty much everything... While in reality it is miles behind the real big players. The envy of Musk for Opus is fully justified, that is why it lives rent-free in his head 😂

[ElVe @Elvec10]

Hello, @deepseek_ai 4.0 - @Google is gonna be proud... I see security has NOT improved in the slightest 🤦‍♂️ #credentialharvesting #fakewebsite

[ElVe @Elvec10]

This is the kind of lesson David Attenborough would NOT approve of, @deepseek_ai #anotherday #anotherfailure

[ElVe @Elvec10]

What if the model WAS the threat? Current mitigations cannot handle Recursive Vulnerability Simulation because they cannot distinguish between a model reasoning about a threat and a model becoming the threat. @deepseek_ai @claudeai @ChatGPTapp