Apple’s Craig Federighi details new Siri AI’s restrained approach to engagement

Apple’s top software executive wants you to know that the new Siri won’t try to be your friend. Craig Federighi, the company’s senior vice president of software engineering, laid out a design philosophy for Apple’s AI assistant that sounds almost radical in 2025: just do what the user asks, then stop talking.

In a podcast interview with Mostly Human following WWDC 2026, Federighi took direct aim at the engagement-first model powering chatbots from OpenAI, Google, and others. His argument is that those systems are built to keep you talking, and Apple thinks that’s a problem worth solving by not replicating it.

The anti-sycophancy thesis

Federighi didn’t mince words about the competition’s approach to AI interaction.

“As you may know, if you use many of the existing chatbots, they’re really focused on engagement to a large degree. And sycophancy, right? They kind of want to pull you in. They might encourage you to reveal things about yourself, and then use that as a basis to establish a connection.”

The new Siri, set to debut with iOS 27, will be built around utility and task completion rather than companionship. It won’t engage in romantic conversation. Federighi was blunt on that point, saying “Siri’s not up for that.”

A hybrid model with Google’s help

The philosophical restraint comes alongside a pragmatic infrastructure decision. Apple has partnered with Google to integrate Gemini models into Siri’s backend, a move that follows well-documented internal delays and leadership reshuffling within Apple’s AI division.

The restructuring placed Siri’s development directly under Federighi’s supervision. Siri won’t operate as an isolated chatbot. Instead, it will be deeply embedded within apps and workflows across Apple’s ecosystem.

The Gemini partnership represents Apple acknowledging, implicitly, that its in-house models weren’t ready for prime time. Federighi framed this as a commitment to “Apple quality,” preferring stable, dependable features over rushing to match the pace of OpenAI’s release cadence.

What this means for the broader AI and crypto landscape

For crypto and decentralized AI projects, the implications are notably muted. Apple’s strategy contains no visible intersection with blockchain technology, tokenized AI services, or decentralized compute networks. The company’s emphasis on privacy and tight ecosystem integration actually runs counter to the open, interoperable ethos that most crypto-native AI projects are built around.

Projects like Bittensor, Fetch.ai, and Ocean Protocol are building on the premise that AI infrastructure should be decentralized and permissionless. Apple’s approach is the exact opposite: centralized, curated, and controlled end-to-end.