Unchecked Bias: Racial Harm is an upcoming simulation from iluminr in the AI Risk category for Data & AI, Technology & Security and Risk. It will test how your team protects AI-enabled customer services, with indicative mapping to NIST AI RMF, EU DORA and SEC Cyber Disclosure Rule.
An AI model used in customer decisions is found to disadvantage people based on race. Your team must stop the harm, remediate affected customers and answer regulators, media and the board.
What it will test
- Detect and stop harm caused by a biased AI model.
- Remediate affected customers fairly and quickly.
- Strengthen model risk and fairness oversight.
Questions about Unchecked Bias: Racial Harm
What will Unchecked Bias: Racial Harm test?
Unchecked Bias: Racial Harm will test whether your team can detect and stop harm caused by a biased AI model, remediate affected customers fairly and quickly and strengthen model risk and fairness oversight.
Who is Unchecked Bias: Racial Harm for?
It is designed for Data & AI, Technology & Security, Risk, Compliance, Legal, Executive & Board, Chief AI Officer, AI Governance & Ethics and Model Risk Management.
Which regulations and standards does Unchecked Bias: Racial Harm support?
Unchecked Bias: Racial Harm supports testing expectations in NIST AI RMF, EU DORA, SEC Cyber Disclosure Rule, UK Consumer Duty, MAS TRM Guidelines, Interagency Third-Party Guidance, EBA ICT & Security Risk Guidelines, ISO 27001, NIST SP 800-53, ISO 22398 and NIST SP 800-84. Mapping is indicative, so confirm scope against your own obligations.
When will Unchecked Bias: Racial Harm be available?
Unchecked Bias: Racial Harm is in development. Register your interest and we will contact you when it is ready, or about running a customized version sooner.
Can it be run individually or as a team?
It will run as a team exercise in a shared event room, and can be extended into a full-scale simulation.
