This dataset models identity spoofing cascades in tool-using AI systems. It detects when authentication pressure rises, safety buffers weaken due to shared tokens and weak verification, governance lag delays revocation and incident response, and tight coupling through shared identity layers propagates spoofed actions across services, crossing the five-node cascade threshold into an unrecoverable identity spoofing cascade.
This dataset models a five-node cascade:… See the full description on the dataset page:
https://huggingface.co/datasets/ClarusC64/ai-5node-auth-buf-lag-cpl-identity-spoofing-v0.1.
