AI Agents
AG-06Draft
Test AI decisions for unfair treatment before use and on a schedule
How Leela would run itEvent-driven · before first use, after material changes, and on the schedule the company sets
Proposed stepsThe requirement
Outcomes of decisions that AI makes or shapes are tested for differences across protected groups before first use, after material changes and on a set schedule, and the method, results, fixes and sign-off are kept. The schedule is a company control, not a legal interval.
- Before this work
- Decision records with outcomes An approved testing method
- Applies
- Tell Leela about your company to see whether this applies.
- Status
- Setup needed
Tell Leela how you handle it today, or hand it over: see what’s next.
How Leela would run it · proposed steps
- LeelaGather the decision records for each round of testing.Proposed step
- LeelaTrack each fix the testing finds until it is closed.Proposed step
- Model risk reviewerA qualified reviewer chooses the method and signs off the results.Your team decides
What Leela keepsThe evidence this obligation leaves, as the work happens
Listed from the policy libraryContents
- 01Fairness testing reportDated report with method, results, fixes and sign-off
Evidence
Fairness testing report
AI Agents §3.6
1
of 1 record
The steps are the library’s proposed workflow for this obligation: a description of the work, not a run. The records are its evidence in the library.