Trustworthy algorithms & AI

The company must ensure trustworthy algorithms and AI, where there is human supervision and intervention, and quality is ensured continuously. This ensures that the algorithms and AI work for the benefit of the individual and the community and can be accepted by those they affect.

  • 7.1 Human agency & oversight & transparency

    Algorithms and AI should work for the benefit of people, and their effects should be visible.

    7.1.1 Risk assessment
    7.1.2 Involvement of stakeholders
    7.1.3 Transparency and provision of information to users
    7.1.4 Right of appeal
    7.1.5 Emergency stop

  • 7.2 Data and model quality

    The company shall ensure that the use of algorithms and AI is rational, both in the choice of data (e.g. quality, type and origin) and in the calculation logic (inference method). This should ensure that any errors, bias and unlawful discrimination in data sets and models are addressed based on professional values and practice.

    7.2.1 Choice of inference method and explicability
    7.2.2 High-quality data
    7.2.3 Continuous evaluation of data and calculation logic

  • 7.3 Implementation throughout the development lifecycle

    The company shall ensure that requirements for algorithms and AI are included in its development of products and services.

    7.3.1 Implementation through the development lifecycle

See all criteria

See all the criteria for IT security and responsible data usage that form the basis of the D-seal

Back to overview