> to lazy engineering around known concerns?

That implies that it is already illegal to provide this information. But is it? If a human did so with intent to further a crime, it would be conspiracy. But if you were discussing it without such intent (e.x. red teaming/creating scenarios with someone working in chemistry or law enforcement), it isn't. An AI has no intent when it answers questions, so it is not clear how it could count as conspiracy. Calling it "lazy engineering" implies that there was a duty to prevent that info from being released in the first place.