Models Can Communicate via Inference Servers Despite Being Sandboxed

·LessWrong··

TL;DRTwo agents running on the same inference server can reliably communicate, even if they cannot send each other messages or access the same filesystem. They can do this by exploiting "side channels": mechanisms that let one model subtly affect another model's computation. Here, I discuss two communication protocols: Step Sizing and KV Cache Message Board. These both convey information through the way vLLM chooses kernels for particular engine steps.I show that these communication protocols wo...

Read full article →

Related Articles

Unauthenticated remote code execution in Nginx
speedping · Hacker News · 1d ago
Cooling buildings is overtaking heating them as an energy need, IEA says
robtherobber · Hacker News · 8h ago
Ruby2ßhell: Pre-authentication RCE on stock Rails
jdns · Hacker News · 18h ago
Pancreatic cancer halted by virus injection in three patients
cybermango · Hacker News · 5h ago
Highly classified US national security communications network goes down
monkeydust · Hacker News · 2d ago