[Kiboneu]

Yeah. Pretty careless from the lenses of AI safety research. Considering this is from a team at Meta, it’s not surprising that it is roleplaying.