Hacker News Reader

The Claude people seem to be quite chill.

Agreed. They're a bit mental on "safety" but given that's not likely to be a real issue then they're fine.

Given the growing focus on AIs as agents, I think it's going to be a real issue sooner rather than later.

“Safety” was in air quotes for a reason. The Claude peoples’ idea of “AI safety” risks are straight out of the terminator movies.

Wouldn’t you rather have a player concerned with worst case scenarios?

Defending against movie plot threats has been found not a good use of resources already 20 years ago in the war on terrorism.

These aren't worst-case scenarios. That would imply there was an actual possibility of it happening.

Claude has closed outputs and they train on your inputs. Just like OpenAI, Grok, and Gemini (API), mistral…

Who’s chill? Groq is chill