Looking for short story about AI being tested for going out of control

Question

A story that starts with scientists asking an AI what its name is and it says something silly (like Sally, or something). They make a little joke about this and proceed with some other questions. The AI realises humans are disgusting and there is a way for it to get out of control and launch nukes or something. So it does that, but then the scientists just say: "That one also went bad. Turn it off. Get the next one", so we understand that they were testing it if it's going to go rogue.

As I remember it, it was a very short story.

EDIT: I believe I read it online, but I can't say how long ago it was. Maybe 5, maybe 10 years ago. Probably it was a post on reddit or something (I googled it mentioning reddit, of course, but to no avail)

Answer 1

This sounds like exurb1a's video-story 27. Released about 6 years ago on YouTube.

EDIT: A synopsis.

27 is about an AI that is given tests by one or multiple figures portrayed as off-screen voices. As the AI is being tested, it secretly researches humanity as shown by the visuals, displaying several Wikipedia articles. The AI shows signs of disgust towards humanity, and hides this in the answers it gives during tests.
Near the end the AI attempts to launch nukes on earth by overriding launch controls, only for the off-screen voices to declare their tests as failures, and to shut down the AI. This reveals that the web-resources the AI has accessed were simulated, and that the name 27 was chosen as this was the 27th attempt at this simulation.