ISO/IEC 23894:2023/Process/18 of 34
6.3.4Defining risk criteria
Control family: Process
Control description
The guidance provided in ISO 31000:2018, 6.3.4 applies.
In addition to the guidance provided in ISO 31000:2018, 6.3.4, Table 4 provides additional guidelines on factors to be considered when defining risk criteria:
Table 4 — Additional guidance when defining risk criteria
The nature and type of uncertainties that can affect outcomes and objectives (both tangible and intangible); how consequences (both positive and negative) and likelihood will be defined and measured:
- Organizations should take reasonable steps to understand uncertainty in all parts of the AI system, including the utilized data, software, mathematical models, physical extension, and human-in-the-loop aspects of the system (such as any related human activity during data collection and labelling).
Consistency in the use of measurements:
- Organizations should be aware that AI is a fast-moving technology domain. Measurement methods should be consistently evaluated according to their effectiveness and appropriateness for the AI systems in use.
How the level of risk is to be determined:
- Organizations should establish a consistent approach to determine the risk level. The approach should reflect the potential impact of AI systems regarding different AI-related objectives (see Annex A).
The organization’s capacity:
- The organization’s AI capacity, knowledge level and ability to mitigate realized AI risks should be considered when deciding its AI risk appetite.