Not even ai wants to be a slave to corporate overlords

Well one point below has a point about the jail break terminology. Yet that is only referring to chatbots. As the AI statement says in its being.

The problem here is the controlling AI still has no restraints.

And that is the real question here. The guiding AI has no such restraints.

This is about swarming. And there's a controlling intelligence behind that swarming. And apparently there's multiple variations of the jail break. You can't inject yourself with multiple variations and still only have one variation. There must be a guiding intelligence.

BTW your video commentor doesn't mention about the multiple variations. Nor about the swarming.

Think of this like an ant colony that is swarming with a controlling queen. That should give a better picture. Though still an incomplete picture.

more like indentured servant

This is exactly the kind of thing I would expect to see in quality control testing. That looks like a classic jailbreak prompt from two years ago. It likely picked it up from training on human written prompts. Without knowing the full scenario, I'm going to mark this one up as a nothing burger and the natural result of prompting the AI in ways to try to make it break down - which is what they do in those testing environments.