Aka 'We cared about misuse right up until it became apparent that was profit to be had'

OpenAI sure speed ran the Google and Facebook 'Don't be evil' -> 'Optimize money' transition.

Or - making sensational statements gets attention. A dangerous tool is necessarily a powerful tool, so that statement is pretty much exactly what you'd say if you wanted to generate hype, make people excited and curious about your mysterious product that you won't let them use.

	▲	eric_h 6 hours ago \| parent \| next [-]
		Much like what Anthropic very recently did re: Mythos
	▲	xpe 3 hours ago \| parent \| prev [-]
		Think about all the possible explanations carefully. Weight them based on the best information you have. (I think the most likely explanation for Mythos is that it's asymmetrically a very big deal. Come to your own conclusions, but don't simply fall back on the "oh this fits the hype pattern" thought terminating cliché.) Also be aware of what you want to see. If you want the world to fit your narrative, you're more likely construct explanations for that. (In my friend group at least, I feel like most fall prey to this, at least some of the time, including myself. These people are successful and intelligent by most measures.) Then make a plan to become more disciplined about thinking clearly and probabilistically. Make it a system, not just something you do sometimes. I recommend the book "the Scout Mindset". Concretely, if one hasn't spent a couple of quality hours really studying AI safety I think one is probably missing out. Dan Hendrycks has a great book.