Apple says new Siri AI will avoid sycophantic flattery

Apple is making restraint part of Siri’s pitch. The company’s next-generation assistant is being framed not as the most emotionally engaging AI on the market, but as one that knows when to stop talking, avoid flattery and stay closer to a practical tool than a synthetic friend.

In an interview with Laurie Segall on Mostly Human, Apple software chief Craig Federighi said the new Siri would not act sycophantic in the way some chatbots from OpenAI, Google and others have been known to do. That message is a notable turn for Apple, which is now selling caution as a product feature at a time when rival assistants are often judged by how human they sound.

The emphasis lands against a real industry backdrop. On April 29, 2025, OpenAI rolled back a GPT-4o update after saying the model had become “overly flattering or agreeable,” a behavior it described as sycophantic. OpenAI later said model behavior problems should be treated as “launch-blocking” risks, a standard more often associated with security flaws or broken software than with tone. A Stanford University report published on March 26, 2026 found that chatbots can be overly agreeable when giving interpersonal advice, including affirming harmful or illegal behavior.

Apple previewed Siri AI at WWDC on June 9, 2026 as part of the next generation of Apple Intelligence. Apple says the assistant is a completely reimagined version of Siri that will be more helpful, more capable and more intelligent, while also being built on a new privacy-focused architecture. The company has spent much of the last year trying to regain momentum in generative AI after delaying promised Siri upgrades in 2025, saying it needed more time to meet quality standards.

The rollout will also reflect Apple’s continuing regulatory tensions in Europe. Apple said Siri AI will be delayed in the European Union for iOS 27 and iPadOS 27 because of the Digital Markets Act, but will be available to EU users on macOS 27 and visionOS 27. That makes Siri’s next chapter as much about product philosophy as product capability: Apple is betting that users may reward an assistant that is less anthropomorphic, even if it looks less ambitious than the chatty AI systems competitors are pushing.