Do you think they currently exist to prioritize AI safety? That shit won’t pay the bills, will it? Then they don’t exist. Goals are nice, OKRs yay, but at the end of the day, we all know the dollar drives everything.
It's simple, they will redefine the term (just like OpenAI redefined "AGI" into "just makes a lot of money) into "doesn't leak user data" and then claim success