I’m truly sorry to be a downer because I’m excited to use Fable again too but (almost) none of y’all actually know what CAISI did and didn’t conclude about the safeguards. And we may not find out anytime soon
#unmuzzleCAISI
I’m truly sorry to be a downer because I’m excited to use Fable again too but (almost) none of y’all actually know what CAISI did and didn’t conclude about the safeguards. And we may not find out anytime soon
#unmuzzleCAISI
No Digg Deeper questions have been answered for this story yet.
I’m sorry to be a downer because I’m excited to use Fable again too but jokes aside, (almost) none of us know what CAISI did and didn’t conclude…
We should still expect a public framework and public evaluations (with appropriate redactions), not just “X agency was involved.”

Update: there's some new info here, though also a lot of just explaining the basics - I still hope we hear more from CAISI over time.
It is a good sign that Ant was able to publish this + say they're involved in work towards an industry standard.
I am def. hoping that in the coming days we will get more public info on these fronts, and the focus now is getting back online then explaining later.
Maybe Ant’s blog tomorrow will elaborate.
But recall the EO called for a classified process. TBD what we will end up getting.
I’m sorry to be a downer because I’m excited to use Fable again too but jokes aside, (almost) none of us know what CAISI did and didn’t conclude…
We should still expect a public framework and public evaluations (with appropriate redactions), not just “X agency was involved.”

@Miles_Brundage Did you read it? Coding/debugging will be nerfed even more. Might not even do it at all.

@Miles_Brundage Doesn't answer everything but this just came out if you haven't seen it https://www.anthropic.com/news/redeploying-fable-5