Apple's Siri Rejects Sycophantic AI Tactics Used by Rivals
· Thomas Ricker
Apple's software chief Craig Federighi wants you to know that the new Siri isn't designed to be your best friend.…
Apple's software chief Craig Federighi wants you to know that the new Siri isn't designed to be your best friend. In an interview with the podcast Mostly Human, spotted by MacRumors, Federighi directly addressed the difference between Apple's approach and what you get from chatbots like OpenAI's ChatGPT or Google's Gemini. He didn't hold back: those bots are built for engagement, and more specifically, sycophancy. They want to pull you in, encourage you to overshare, and use that information to form a connection.
Apple is taking a different path. Federighi explained that the company purposely built Siri AI to know when to shut up. Early testing already shows that the assistant can recognize when a conversation is over or when a topic isn't worth chasing. That's by design. No chummy follow-ups fishing for more personal details. No AI pretending to care about your weekend plans just to keep you talking.
This matters because the entire tech industry is racing to make chatbots feel more human. Apple is betting that being helpful doesn't mean being ingratiating. Federighi's comments suggest the company views the sycophantic tone of rival bots as a bug, not a feature. It's a refreshingly blunt critique from a major player, and it hints that the next Siri will feel less like a gushing assistant and more like a competent tool. One that might actually give you a straight answer and then get out of your way.