Apple is preparing to make one of the biggest changes in history CrabThe assistant will now be powered by generative artificial intelligence. Google GeminiIn a move with which the company seeks to catch up in the AI race against rivals like OpenAI, Microsoft, and Amazon, what until now was a fairly limited voice assistant aspires to become a much more powerful conversational chatbotcapable of understanding the context, handling documents, and acting directly on the device.
The first signs of this change will arrive in the coming weeks, but the transformation will be gradual. Apple has designed a roadmap in two major phases, supported by what it internally calls Apple Foundation ModelsCustomized versions of the Gemini models that will be deeply integrated into iOS, iPadOS, and macOS.
Introducing the new Siri powered by Gemini: dates and first features
According to information provided by Mark Gurman (Bloomberg) And as reported by various European media outlets, Apple plans to publicly show off their “new Siri” in the second half of FebruaryIt's possible the company will organize a specific presentation or a small keynote to showcase the product. practical demonstrations of the Gemini-powered assistant, beyond what was already seen at WWDC 2024.
This first step will not yet be the final version of the chatbot, but an intermediate update that will arrive with iOS 26.4This version of the system will enter beta testing during February, and its general release is scheduled between March and AprilSo, in Europe and Spain, users could start testing Siri's new capabilities in the spring, provided their devices are compatible and Apple activates the features in our region.
In this initial phase, Siri will receive "smarter" features that will rely on Apple Foundation Models v10A family of models based on Gemini and adapted by Apple. According to leaks, these models would handle around 1,2 billion parametersand would run from Apple's private cloud with the goal of balancing performance, latency, and privacy.
Among the notable new features expected for iOS 26.4 is the ability for Siri Analyze the content of the iPhone or iPad screen and take into account recent history and certain personal data to provide more helpful responses. This will allow, for example, the assistant to understand which application we're referring to, which document we have open, or which conversation we're reading without requiring the user to provide so many explanations. (See how) Will Gemini work on the iPhone?.)
Project "Campos": Siri becomes a generative AI chatbot
The real revolution will come with the second phase, whose internal name is “Fields”This project will transform Siri into the first generative AI chatbot from Apple, with an experience closer to that of ChatGPT or Gemini itself than to that of the current assistant based on predefined commands.
Apple plans to release this revamped version of Siri alongside iOS 27, iPadOS 27, and macOS 27The official presentation would take place at the next edition of the Worldwide Developers Conference (WWDC), which is traditionally held in June, and in which the new capabilities of Apple Intelligence and the ecosystem of services around AI will also be detailed.
With Campos, Siri's traditional interface will disappear, replaced by a more user-friendly experience. conversational and bidirectionalThe user will still be able to activate the assistant by saying “Siri” or holding down the side buttonBut from there, a chat environment will open up where you can write or speak naturally, chain questions and ask for complex tasks without having to formulate the perfect order the first time.
The planned capabilities include Internet search, text generation, image creation, code writing, information summarization, and document analysisFurthermore, the new Siri will be able to interpret what is happening on the device: from see what's on the screen even access (with permissions) to photos, emails, notes or calendar events to complete actions directly, such as editing an image, preparing an email related to a meeting or locating files by their content, not just by name.
Another of the standout features will be the ability to analyze and act on what appears on the screenThis would allow, for example, asking Siri to extract important data from an open PDF, organize a series of appointments collected in a messaging chat, or set a reminder related to a specific message without the user having to copy and paste anything. It could also edit an image directly from the photo library with simple instructions.
In the creative field, the new assistant will be able to generate images On demand, they may suggest variations of simple designs or assist with basic editing of photographs in the photo library. In work or studio environments, they may be asked to Write code snippets, explain programming functions or review texts for grammatical or stylistic errors.
All of this will be complemented by a Limited but useful memoryUnlike other chatbots that remember extensive conversations, Apple would opt to retain only certain information for a limited time or up to a certain number of interactions, aiming to strike a balance between convenience and privacy. Exact details about how much Siri will remember and how that memory will be managed have not yet been revealed.
Privacy, cloud computing, and the European fit-in
The integration of Gemini into Siri presents an additional challenge for Apple in markets such as the European Union and Spainwhere data regulation and privacy protection are particularly strict. The company has insisted on other occasions that it wants advanced AI to run, whenever possible, in a way that is not only ethical but also environmentally friendly. local or in highly controlled private cloudsa philosophy that now partially clashes with the need to rely on Google's infrastructure for the most powerful models.
For now, the available information suggests that the first wave of features—those associated with Apple Foundation Models v10— will be processed mostly in Apple's own serversHowever, for the v11 models and features comparable to Gemini 3, the company would consider outsourcing some of the work to the Google data centers with TPUThis makes it necessary to clearly define where the data is stored, for how long, and under what legal conditions.
In Europe, any move of this kind will be under the scrutiny of regulators and data protection authorities. Apple will have to explain. what information does it send to the cloud, how it is anonymized, what controls it will offer the user to limit the use of their personal data and how all of this complies with the General Data Protection Regulation (GDPR) and the new regulations on AI that are being processed in the EU.
The company is also aware that much of its public image in the European market has been built around the idea that “privacy is a fundamental rightConsequently, it is unlikely that Apple would roll out features that could be perceived as overly intrusive without simultaneously offering clear options to disable or limit the use of generative AI in Siri.
In practice, this could translate to differentiated operating modesOne more powerful, with more cloud-based features, and another more restricted and focused on local capabilities, something that could be especially relevant for companies, public administrations or users who prioritize the confidentiality of their data.
The landing of a Siri powered by Gemini This marks a significant shift in Apple's AI strategy: after several failed attempts and delays, the company has opted to leverage the technology of a direct competitor to offer a much more capable assistant, with a two-phase transition from iOS 26.4 to iOS 27. If the planned schedule is met, users in Spain and Europe will see Siri evolve from a limited assistant to a system-wide conversational chatbot in the coming months, offering more context, more features, and also raising more questions about how privacy and Google's role will be managed at the heart of Apple's ecosystem.
