Apple plans a major overhaul of Siri later this year, turning the voice assistant into a conversational artificial intelligence chatbot and pushing the iPhone maker into more direct competition with OpenAI and Google in generative artificial intelligence (AI).
The redesigned system, internally named Campos, will, according to a Bloomberg report, replace Siri’s existing interface and embed deeply across the iPhone, iPad and Mac, moving the assistant beyond simple commands towards a more advanced, chatbot-driven artificial intelligence experience.
Campos will launch the same way users currently summon Siri but enable fluid, two-way conversations similar to ChatGPT or Gemini, overcoming long-standing limits that have kept Siri confined to basic, task-focused interactions.
Apple plans a two-stage upgrade for Siri, starting with iOS 26.4 improvements to existing features, followed later by a full chatbot rollout that will mark a much bigger shift in how the assistant works.
The initial update will add abilities such as analysing on-screen content, using personal data like messages and calendars and improving web search, while keeping Siri’s familiar interface instead of introducing a chatbot right away.
Apple plans to build Campos directly into iOS 27, iPadOS 27 and macOS 27, prioritising system stability, performance and artificial intelligence capabilities over major visual changes across its operating systems.
Campos will likely gain deeper access to system functions, allowing it to control settings, open apps, edit photos, draft emails, manage calls and interact smoothly with Apple services including Mail, Photos, Music and Podcasts.
The chatbot initiative reflects Apple’s attempt to rebuild momentum after an uneven Apple Intelligence rollout in 2024, as competitors such as Samsung and Chinese smartphone brands accelerate the integration of conversational artificial intelligence.
Despite highlighting in-house development, Apple may, according to reports, depend on Google artificial intelligence models for Campos, potentially hosting the system on Google servers while paying about $1 billion annually for access to advanced models.
