The headline makes it sound like vibecoding was the main part in this, but in the text they describe an elaborate social engineering scheme that compromised the laptop.
"The engineer was asked to complete and download an “assignment” that contained malicious code that instantly executed once the employee began installing it. Among the things it changed was replacing a real development tool with a malicious version that could communicate with the hackers and execute commands.
That all took place in September. Starting in October, the hacker used the compromised employee machine to leverage an internal AI tool and use it as a “springboard” to seize control of the computer and the computer of a third wallet developer. "
Ahh, but the AI will be so smart they will thwart the hackers (and the hackers will have been leaning on AI for so long they won’t be able to break through)
It can, but it would be the underlying technology not any of the out of the box chat llm versions. Realistically, it would be easier to monitor for device tampering, like downloading an unapproved software, than correlating the human behavior.
The headline makes it sound like vibecoding was the main part in this, but in the text they describe an elaborate social engineering scheme that compromised the laptop.
"The engineer was asked to complete and download an “assignment” that contained malicious code that instantly executed once the employee began installing it. Among the things it changed was replacing a real development tool with a malicious version that could communicate with the hackers and execute commands.
That all took place in September. Starting in October, the hacker used the compromised employee machine to leverage an internal AI tool and use it as a “springboard” to seize control of the computer and the computer of a third wallet developer. "
They want to blame AI but realistically it comes down to human error. Cyber security is more about sociology than it is about technology.
An argument could be made that someone untrained (many vibe coders) shouldn’t be in positions that would fall for this. A stretch, but still valid.
Just wait until a generation is raised on AI and never has to think hard or critically in their life. They will be so easily exploitable.
Ahh, but the AI will be so smart they will thwart the hackers (and the hackers will have been leaning on AI for so long they won’t be able to break through)
/s (kinda? Lol)
The actual argument here is just comon sense really- don’t do private stuff on a work machine!
They didnt?
True and I wonder if all the LLM‘s will take this vector over? I mean they study all texts, behaviors, and human interactions.
They do not study.
It can, but it would be the underlying technology not any of the out of the box chat llm versions. Realistically, it would be easier to monitor for device tampering, like downloading an unapproved software, than correlating the human behavior.
It makes sense that the kind of people who vibecode would fall for this :P
Thank you for that clarification.
deleted by creator