Compliancemanzill.com
ISO/IEC 23894:2023

ISO/IEC 23894:2023/Process/25 of 34

6.4.2.6Identification of consequences

Control family: Process

Control description

As part of AI risk assessment, the organization should identify risk sources, events or outcomes that can lead to risks. It should also identify any consequences to the organization itself, to individuals, communities, groups and societies. Organizations should take particular care to identify any differences between the groups who experience the benefits of the technology and the groups who experience negative consequences.

Consequences to the organization necessarily differ from those to individuals and to societies.

Consequences to organizations can include but are not limited to:

- investigation and repair time;

- (work) time gained and lost;

- opportunities gained or lost;

- threats to health or safety of individuals;

- financial costs of specific skills to repair the damage;

- employee recruitment, satisfaction and retention;

- image reputation and goodwill;

- penalties and fines;

- customer litigations.

Depending on the context, consequences to individuals and to societies can be more general, in which case the organization can be unable to estimate exactly what the effect to each individual or to societies is.

Rather than specifying each type of effect, this can be considered generally as with the degree of the criticality of effects (for example, to privacy, fairness, human rights, etc., in the case of an individual, or to the environment in the case of societies) being a key element.

The exact effects can depend on the context in which the organization operates and areas for which the AI system is developed and used.

How to implement

Coming in phase 2Step-by-step guidance for implementing this control on Microsoft Azure, using services such as Azure Policy, Microsoft Defender for Cloud and Entra ID. It will be matched to your environment.