Include mitigation measures to address biased output during the development of artificial intelligence systems.

CONTROL ID
15047

CONTROL TYPE
Systems Design, Build, and Implementation

CLASSIFICATION
Corrective

This Control directly supports the implied Control(s):

Establish, implement, and maintain an artificial intelligence system., CC ID: 14943

There are no implementation support Controls.

High-risk AI systems that continue to learn after being placed on the market or put into service shall be developed in such a way to ensure that possibly biased outputs due to outputs used as an input for future operations ('feedback loops') are duly addressed with appropriate mitigation measures. (Article 15 3. ¶ 3, Proposal for a Regulation of The European Parliament and of The Council Laying Down Harmonized Rules On Artificial Intelligence (Artificial Intelligence Act) and Ameding Certain Union Legislative Acts)
potential bias impact to an individual; (§ 6.4.3.2 ¶ 3 Bullet 3, ISO/IEC 23894:2023, Information technology â Artificial intelligence â Guidance on risk management)
protections and mitigating controls around unwanted bias and unfairness; (§ 6.4.3.2 ¶ 3 Bullet 7, ISO/IEC 23894:2023, Information technology â Artificial intelligence â Guidance on risk management)