Hacker News new | past | comments | ask | show | jobs | submit
I know that this comment will fall on deaf ears here on HN, but the reason Amodei doesn’t like open-source is because he’s worried that as intelligence scales, people will be able to use future abliterated open source models to perform large scale hacking or bioterrorism. In my mind, this is a perfectly reasonable concern, and frankly I’m very worried about a future where someone can abliterate GLM 7 or whatever and then just have a swarm of world class hackers they can point at anything or anyone.

You can disagree with him, but Amodei and most of Anthropic are true believers in superintelligence and what that will mean for humanity. Everything they do makes way more sense if you look at their behavior through the lens of being genuinely concerned about AI safety.