Again, not that I have seen any indication to suggest we're anywhere close to any of that.

But in a bullshit-supersaturated environment, this part seems even more bullshit-y than most.

You have all the public posturing about "oh we take AI alignment and ethics very seriously" and then you see things like the Claude code leak and, as expected, the reality is mostly prefixing your prompt "Please be a nice and fluffy computer, not nasty, uwu." and hoping for the best