nothing obvious comes to mind and I am not aware of any well-known bug looking like this.
Do you graph this CPU OIDs? if yes, you can look at the graphed value at the time the alert is sent with a value of 7 or 9. Does the chart show CPU value around 7/9 or 75%
Is it the first advanced alert on UnDP value, that you configure? Do others work well?
If nothing like this sheds any light or there isn't any mis-configuration, I suggest open a case.
yesterday i did snmp map the oid out and the cpu leve was fine, so there looks to be a little bug with the advance alert