AI is not sentient, therefore not liable for its actions. It would be pointless. We ought to look one step beyond the lines of code for liability.
Unless the AI eventually passes the bar of sentience.
Unless the AI eventually passes the bar of sentience.
I think the responsibility should lie on the shoulders of the individual/business that deployed the AI agent. For that, I can clearly see the benefits to society.
... and subject to meaningful punishment.
Not sure I can visualise that.