Once you enter the world of operating outside IC specifications, you are inviting in shaky, unstable behaviour and odd faults that take time to track down.
You may see a cost saving in using a chip outside its specification. That saving may seem attractive or worthwhile because the proper chip is much more expensive, the board would need rework to use the proper chip, the chip would need extra interface components etc. That cost may be multiplied by the number of 'wrong' boards you have made and in stock.
That must be traded off against the cost of testing and proving sufficient boards to be sure it will work and the cost of tracking down obscure faults that might result from it. And most of all, the possible costs of recalling and replacing this dodgy design with the corrected boards that you avoided making in the first place and that have to be made after all.
Its worth noting that reliability faults or errors through instabilities will not reveal themselves until you have a large-enough number of units in use (read up on MTBFs). That may mean one unit before the fault shows, it may mean a hundred or a thousand - it depends upon the components and the design.
So the reason why going outside IC spec' is a bad idea may not become apparent until you have made and used lots, when the corrective action costs have got higher. So just running a few in a lab' under test for a bit cannot to be taken as definitive proof of reliability.
As the going-wrong costs usually considerably outweigh the cost savings if taking a chance goes right, then I would never use ICs outside of their rated specifications.
You know your circumstances but I recommend that you do not. Select the correct IC instead or add a voltage translation circuit to get the right voltages in/out of that IC.