Lasq📣 NEW VIDEO ALERT!
I've resurrected my YouTube channel to dive into Prompt Lock ransomware. Discovered by ESET and branded as "first known AI-powered ransomware.", it caused a lot of buzz in the media earlier this year.
Even though this was exposed just as a Proof-of-Concept, I was curious how it would perform in the lab environment. I didn't see anyone actually testing this malware before so, I've done just that. Does it live up to the hype? Is AI-driven ransomware the future of threat?
Spoiler alert: It failed miserably. 🤯
In this video, we go deep into the reverse engineering:
Setting up the lab: Using LMStudio and Burp Suite to successfully proxy the malware's Ollama API calls to a local LLM server.
Watching it run: We analyze the verbose Lua script generation chain in real-time.
The Flop: We uncover the hilarious and critical failure points, including massive LLM hallucinations where it invented sensitive files (Resumes, Bank Statements, Medical Records) that didn't exist, and completely botched the final ransom note!
This highlights the critical limitations of integrating LLMs into live exploit chains.
Watch the full breakdown and the spectacular failure here: https://www.youtube.com/watch?v=-qex_aqN3LA
#Cybersecurity #Ransomware #AI #LLM #MalwareAnalysis #ReverseEngineering #PromptLock #ThreatIntelligence #MalfindLabs
Kevin Beaumont
Mateusz Chrobok
Pascal Leinert
jpmellojr
ESET Research
ccinfo.nl
nickbalancom
sekurak News