Anthropic releases post on AI threat assessments
Anthropic released a post on AI threat assessments while negotiations on international coordination for the technology remained active. The timing drew attention to possible alignment with those talks. Observers noted that the positions match statements Anthropic has made in prior public updates and social media posts. Discussion separated questions of substantive disagreement on threat evaluations from separate issues around the choice of release timing.
Maybe it’s something they’ve been saying for a long time now? My point is it shouldn’t matter. One should disagree on substance if you want (and there are some things reasonable people can disagree on) but nothing in that post is something they haven’t been saying very publicly (including in public posts) for a long time.
Shakeel@ShakeelHashim@dkaushik96 @michaelsobolik What I’m trying to say is that this post reads as an effort to actively *hamper* the talks. That appears to be why it was published. It is hard to see another reason why they would publish it and especially publish it today. I don’t get why Anthropic would do that.