Protecting Proprietary Weights

Protecting Proprietary Weights

Flag agentic leaks of trading models, customer PII, and proprietary weights at the ioctl boundary.

Imagine a quant desk training proprietary alpha models on H100s. The risk: a contractor's notebook fires a large ioctl that pulls weights off the device. The hot-path heuristic flags transfers above the configured byte threshold; a sample is forwarded to the on-host SLM, which scores risk 0–100 and returns a reason. Above threshold, the daemon writes a structured alert and (if mode = kill) SIGKILLs the offending PID before the next ioctl batch lands. No data leaves the host at any point.