There are ongoing ways to keep an eye on known AI risks as time passes, so a small problem does not quietly grow worse.
Think of it likeKeeping a chart on the wall for a slow leak under the sink, checking it weekly, so you notice if the drip is getting worse before the floor rots.
In plain English
Set up a way to keep watching the AI risks you already know about, so you catch a problem that is slowly getting bigger before it becomes a crisis.