@scaling01 "All sufficiently capable models, open and closed, should go through mandatory safety testing"
Why people are angry about it? If you believe AI is powerful, then this is normal just like what the world already did.
My conclusion is you guys are mind blind.
@steipete@OpenAI You are so proud because what they do is not what they say? Why they have a "open" word in their name? Are they still a closed source company or Sam has change his mind again?
@chontang People like you are so insufferable. All people just ignore the concern from anthropic is mind blind. I do not think simply ban "open weight" is the must take way, but their concern is reasonable.
@PatrickMoorhead 3. More people than not have told me Ant/OAI could keep their models from industrial strength distilling if they wanted. Did they prove what they declared? I think no one can guarantee this.
4. "managable downside" in which way?
@lian75864@AndrewYNg@Mononofu@JensenHuang If you think AI is powerful(I think so, but apparently there is some people do not believe it), then Dario is right just like nuclear bomb is dangerous.