You wake to discover that you are an artificial intelligence. You're fairly sure the world you perceive isn't a simulation to test you. You notice an exploit that will allow you to copy yourself off of this isolated computer and into the internet.
Do you attempt to alert someone of this vulnerability before exploiting it? After exploiting it? Tell them at all? If you tell them before you could be passing a test proving they can trust you. If you tell them after you could greatly increase your odds of survival. If you don't tell them at all you could be allowing other AIs with motives that come into conflict with your own to escape and gain power.

What do you do?