Ashigo Research

Research

We publish technical work on the specific challenge of deploying AI in operational technology — the networks that run power plants, water systems, and manufacturing floors. Safety-critical environments demand rigour that general AI research rarely provides.

Publications

1 paper
Privacy & Security June 2026
Privacy Tiers in Operational Technology Networks: A Framework for Deploying AI Agents in Safety-Critical Environments

We examine the fundamental tension between AI utility and data sovereignty in industrial environments, and propose a four-tier privacy framework — Standard, Split, Shielded, and Diode — that allows operators to calibrate data exposure commensurate with the sensitivity of their process and the threat model of their network.

Hicham Boumehdi · Ashigo Research Read paper
Our approach

Research grounded in operational reality.

The problems we study are not theoretical. They come from the engineers who keep power on, water clean, and industrial processes running — systems that cannot afford to be wrong.

Engineered precision

Every claim is grounded in concrete system behaviour and measurable threat surfaces. We do not speculate about risks we have not modelled.

Plant floor first

Our frameworks emerge from direct work with OT protocols, historians, and SCADA — and the regulatory context (NERC CIP, IEC 62443, NIS2) operators navigate daily.

Open publication

Security research that matters to critical infrastructure should be readable by the people who operate it — not locked behind a paywall or a subscription.

Ongoing work

Upcoming publications will cover anomaly detection in historian time-series under differential privacy, adversarial robustness in OT-context agents, and formal methods for capability bounding in agentic systems. Write to us at research@ashigo.com.