Hacker News Reader

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.