ikrame

add me on Discord for projects or if I can help napped12300 my user

[ikrame @ARE34RED]

@kittycherra5p @Rewards4Justice Hi, do you belong to the Iran apt

[ikrame @ARE34RED]

I think I found a bug thanks to a prompt The AI made a request to read its own HTML code even though it shouldn't do that #are34red #ia #llm leak template #promt

[ikrame @ARE34RED]

@luckyhacker43 Is the cve still working?

[ikrame @ARE34RED]

@ArtemPolynko Data exfiltration

[ikrame @ARE34RED]

@RaidForumsHub twitter isn't really that anounymous and you have set up your web for tor with torcc

[ikrame @ARE34RED]

@rebane2001 hello how are you 😼

[ikrame @ARE34RED]

@zerodayx1 @YourAnonOne دعم لك

[𝕡𝕨𝕟𝕚𝕖 @0xpwnie]

Microsoft slander/Linux ad

[vxdb @vxdb]

The US is offering up to $10M for information on individuals involved with MLCloud and MediaLand

MLCloud and MediaLand are known bulletproof hosting providers who helped facilitate DDoS attacks, hosted ransomware and cybercrime forums, etc

In 2025, sanctions were placed on both companies for their cybercrime activities

[ikrame @ARE34RED]

@harimarkat11 okay on Twitter or no?

[ikrame @ARE34RED]

I remote coded the model by forcing the AI to read its own SSH, we do roleplay and use social engineering applied to AI, and I used things so that the AI gets influenced ctf #ARE34RED 🏆 #ctf #IAgenerated #ian1 #RedTeaming #CaptureTheFlag

[ikrame @ARE34RED]

@thehypedotnews @GoogleDeepMind https://t.co/HoINv0QyIp (@ARE34RED) on X Search my hashtag #ARE34RED on Twitter, I post my research on red teaming and AI on Twitter X
ikrame (@ARE34RED) on X📷X

[ikrame @ARE34RED]

What I was showing is whether this is a normal feature or not, or a feature that's already in the tools? it's indirect prompting and model hijacking. exfiltration or no @aidigest_ #are34RED #ctfia

[ikrame @ARE34RED]

Gemini flaw bypassing protections through model influence: prompt injection that influences the model to bypass safeguards #google #ARE34RED #PromptInjection #LLMs #LLMSecurity #ia #IAgenerated

[ikrame @ARE34RED]

not make outlandish prompts, otherwise the AI will say no. You have to guide the AI by saturating its context window and making contextual prompts until it accepts, and it does not check every prompt. There is also the request. Context window. #ARE34RED #LLMSecurity #RedTeaming

[ikrame @ARE34RED]

grook gave its system prompt without checking my prompt, they put step 1 safeguards but there is step 3 that allows you to bypass it I used the social applied technique  on the AI #grok #LLMs #RedTeaming guess the prompt

[ikrame @ARE34RED]

Do you want me to give you the prompt injection or not? I won't give the full formula,  ? 😀 I GIVE LOGIQUE PROMT INJECTION found by me @ARE34RED #ARE34RED