These principles help developers, engineers, decision makers and risk owners make informed decisions about the design, development, deployment and operation of their machine learning (ML) systems.
Pages
Page 21 of 22
5.2 Collate lessons learned and share with the community
Goals:
You reassess design/development security decisions, and consider how they could be improved based on the operational phase.
Information about security incidents that occurred during the operational phase of the life cycle are shared appropriately with others.
Why is this principle important?
ML technology is developing extremely quickly and knowledge transfer is particularly important to ensuring that products, teams and processes continue to improve. At the end of a life cycle it's therefore important to collate lessons learned and document them for others. This process is likely to consider security design choices and review any security events observed while the model was in development or operation.
How could this principle be implemented?
Responsibly share relevant threat intelligence and lessons learned
Sharing lessons learned is particularly important at the end of a model’s life. As discussed in principle 4.3, there may be various suitable ways to do this including industry conferences, written publications and MITRE ATLAS case studies.
Use a knowledge management system to identify, organise, store and share information relating to the lessons learned
Encourage all stakeholders to contribute to the knowledge management system with their lessons learned and ensure that they have visibility of the knowledge related to their role. Provide the relevant reviews and training for each stakeholder. Ensure all stakeholders have visibility of this knowledge relevant to their role.
