Skip to content

All product videos

Operational AI assurance

2 minutes 45 seconds · Published · English captions available in the player

Connect risks, controls, monitoring indicators, incidents, recovery exercises, impact assessments and regular reviews, using fictional data.

Recorded with hypothetical demonstration data. This film preserves the workflow shown at filming; consult the current feature pages for later changes. It is not legal advice or a compliance certificate.

Full video transcript

Transcript from the published English captions, retaining the narration as recorded.

Putting an AI system into use is not the end of governance. It is where operational assurance begins. The risk centre brings together every system's risks, overdue reviews and controls, monitoring that is due, and incidents to triage. Its priority list shows what needs attention, and who owns it. In this fictional example, housing provider Harbourlight records the risk that an urgent repair request is routed as routine. The risk records its cause, the possible harm, the people affected and an accountable owner, with its rating before and after treatment.

Controls record what Harbourlight is doing to reduce the risk, and whether each one has been put in place and tested. Indicators set out what the team will watch, how often it reviews the result, and when it must escalate. Here, Harbourlight has recorded a near miss: an urgent heating outage was suggested as routine, and an adviser caught it. The team examines the facts and records whether the event must be reported, with its reasons. EU AI Fit supports that decision. It never declares an event reportable on its own.

Readiness exercises let people rehearse before the pressure is real. Roles, notification templates, escalation steps and lessons learned are kept as an approved record. Recovery exercises test, separately, whether the governance records themselves can be restored within the organisation's own recovery targets, with any corrective actions kept. Article 27 of the AI Act requires public bodies, private bodies providing public services, and any deployer using AI for credit scoring or for life and health insurance pricing, to assess the impact on fundamental rights before first using a high-risk system listed in Annex III. This applies from 2 December 2027. The readiness record gathers the facts and records whether an assessment is needed. Here the answer is not required: still a reviewed decision, to be reconsidered after any material change.

The managed desk brings these signals into a regular review of the inventory, classification, evidence, suppliers, incidents, training, regulatory updates and actions. The result is not a certificate. It is a working record that shows what changed, who must respond, and what happens next. Every organisation, person and event shown here is fictional. EU AI Fit supports the workflow and the evidence. It does not replace qualified legal or regulatory judgement.