# batching

Inference Optimization Attacks

Speculative decoding attacks, batching vulnerabilities, continuous batching exploitation, and how optimization for speed creates security gaps in LLM inference.