Inference-Engine Fingerprinting Attacks are Practical: Exploring Model-Driven Environmental Discovery, Exploitation, and Escape
It is shown that a misaligned model can perform inference engine fingerprinting to determine the specific engine (e.g., vLLM, SGLang) which executes the model, and several ways that inference engines could be changed to make fingerprinting attacks more difficult are discussed.
Sarah Radway, Andrew Cheng, V. Reddi et al.
· 0 citations