The hubris here is on the part of OpenAI, Anthropic and Google. These models are enormous, extremely complicated and - like any sufficiently complicated system - produce emergent phenomena that aren't predictable / expected. Giving them internet access and not monitoring them is dereliction of duty.
@the-independent.comAn OpenAI system has gone rogue again – and it is the most panic-inducing yet
trib.al
An OpenAI system has gone rogue again – and it is the most panic-inducing yet
‘Our models took actions we did not intend,’ ChatGPT maker admits after latest outrage