Delay the launch until conformity requirements are: What is the most defensible release-readiness decision?
For a high-risk system, release readiness requires satisfying conformity requirements and completing the model card before going live.
The question
A high-risk AI system is scheduled to go live next week, but its conformity assessment is incomplete and its model card is only partially drafted. Business leaders are pressing to launch on schedule to meet a customer commitment. What is the most defensible release-readiness decision?
Preparing for AIGP? Take the free 5-min readiness quiz →
- Delay the launch until conformity requirements are satisfied and the model card is complete, since release readiness depends on both being in place. ✓Correct: for a high-risk system, readiness requires completing conformity requirements and release documentation before production, so the launch should wait.
- Launch on schedule under a conditional exception and finish the conformity assessment and model card within the first month of live production operation.Plausible but wrong: releasing a high-risk system before conformity is satisfied inverts the readiness gate and exposes the organization to real risk.
- Reclassify the system to a lower risk tier so the outstanding conformity steps no longer apply and the planned launch date can be met.Plausible but wrong: risk classification follows the system's actual use, not the schedule; reclassifying to dodge obligations is not defensible.
- Launch to a limited pilot group without the conformity assessment, treating the model card as optional documentation for an early release.Plausible but wrong: a narrower audience does not remove conformity obligations for a high-risk system, and the model card is a readiness requirement.
The trap
Letting a deadline justify releasing a high-risk system before conformity and documentation gates are met. How to remember it
For a high-risk system, release readiness requires satisfying conformity requirements and completing the model card before going live.
How many of these would you get right?
One of 1581 AIGP questions on Certsqill. Take a free five-minute check and see your score per domain — not one number, but which section to open tonight.
Test your AIGP readiness — freeMore Understanding How to Govern AI Development questions
- Recording datasets: Which practice best satisfies all three needs at once? →
- Treat the input shift as data drift and trigger scheduled: Which maintenance action does this pattern most →
- Red teaming that deliberately stress-tests the system: Which activity best fits this specific goal? →
- All 426 Understanding How to Govern AI Development questions →
Part of the Certsqill AIGP question bank · Understanding How to Govern AI Development ·
Every answer, right and wrong, comes with its own explanation.