EvidenceSheet

MN-3.2 Pre-trained models which are used for development are monitored as part of AI system regular monitoring and maintenance

Pre-trained models which are used for development are monitored as part of AI system regular monitoring and maintenance. Pre-trained and transfer-learned models are treated as a monitored component in their own right, si

4
artefacts
1
held by a system
1
at each review
moderate
to go live
SIEM / log platform
where the evidence lives
teal = a system already holds it · olive = produced at each review

system holds itEvidence a system already holds

  • Monitoring records covering those models within regular maintenance · SIEM / log platform

periodic reviewEvidence produced at each review

  • Assessment of risks carried over from the pre-training data and objective · HR system / LMS

governing documentDocuments that govern the control

  • Identification of pre-trained models used, with version and provenance · Vendor register / contract repository
  • The procedure followed when the upstream model is updated or withdrawn · Vendor register / contract repository

First move

Start with the 1 of 4 artefacts that already live in a system (SIEM / log platform); keep the periodic reviews but log each one as a dated record with a named reviewer.

Common gaps auditors find

Do this for your whole sheet

Paste the rows you run your controls from and get this mapping for every control at once, with the periodic-review ones flagged and a first move per row. No account for the first run.

Build my evidence sheet

MN-3.1 AI risks and benefits from third-party resources are regularly monitored, and risk controls are applied and documented · MN-4.1 Post-deployment AI system monitoring plans are implemented, including mechanisms for capturing and evaluating input from users and other relevant AI actors, appeal and override, decommissioning, incident response, recovery, and change management