AI is now a NERC CIP control you have to evidence.
Tokto gives the utility or port risk and compliance officer one record that ties every prompt, every model output, and every vendor-shared AI use to an asset, a facility, and an operating state, ready for NERC, FERC, TSA, CISA, MARSEC, and the board risk committee.
A NERC audit asks for the AI control evidence behind a CIP requirement. CISA wants to know what an AI tool touched before an incident. The board risk committee asks where AI sits relative to the OT risk appetite. The risk function has OT logs, IT logs, and a vendor register that never connect a model output to an asset and an operating state.
- Every AI interaction scored and recorded against an asset, a facility, an operator, a model version, and the operating state in force.
- A single evidence layer that NERC, FERC, TSA, CISA, MARSEC, and the board risk committee read against the same record.
- Policy enforced at the prompt: grid telemetry, port manifests, and ICS commands blocked before tokens leave the boundary.
- AI risk that is measured against the OT risk appetite, examinable under CIP, and attestable to the board.
- AI never enters the OT risk register. It surfaces first as a CIP violation at up to seven figures per day.
- A CIP-013-2 supplier event runs through an AI co-pilot. The audit trail is missing the AI layer.
- An operator pastes grid topology into a public model. The exposure is found after the fact, not on the register.
- The insurer asks how AI risk is controlled at renewal. There is no attestable answer. The premium reprices.
Tokto turns AI on the IT/OT seam into a managed, evidenced control. Every control-room co-pilot, every field assistant, every vendor-shared model becomes a scored record at the moment of output, tied to the asset, the facility, the operating state, and the policy in force. The risk function finally has the AI layer the CIP standards assume exists.
When NERC examines a CIP control, when CISA asks what the AI did before an incident, when MARSEC asks how port manifests were governed, the answer is one query against the system of record. The CRO reports AI exposure alongside the rest of the critical-infrastructure risk register, with a control and a trail.