My guess, the subject matter likely triggers a safety protocol. How often does the bot cite sources in that way? It’s not a conversational reply that’s trying to emulate a human.