Tag: #ml

Researchers Find Position-Independent KV Cache Reuse Can Let One User Hijack Another’s LLM Output

Researchers show position-independent KV-cache reuse in LLM inference can let attackers' context influence other users' outputs, posing risks for shared serving.

#ai, #security, #infrastructure, #ml, #llm

Oxford preprint finds GraphCast and NeuralGCM fail basic coordinate-change tests

An Oxford preprint on arXiv reports GraphCast and NeuralGCM fail latitude/longitude flips and rotations, raising questions about physical generalization.

#ai, #weather, #climate, #ml