Visual intelligence—the phone learning to see and reason about what it sees
En junio de 2026, Apple presentó iOS 27 en su conferencia anual para desarrolladores, marcando un giro filosófico profundo: la compañía que durante años construyó su identidad sobre la autosuficiencia tecnológica y la privacidad ahora se apoya en los modelos de inteligencia artificial de Google para reimaginar a Siri. Esta alianza no es solo técnica; es una declaración sobre los límites del individualismo corporativo frente a la velocidad del cambio tecnológico, y sobre cómo la privacidad puede coexistir —o no— con la dependencia de infraestructuras externas.
- Apple rompió con años de desarrollo propio al integrar los modelos Gemini de Google en el corazón de Siri AI, una decisión que sacude la narrativa de independencia tecnológica que la empresa había cultivado cuidadosamente.
- La tensión entre potencia y privacidad se resuelve mediante un sistema híbrido: las tareas simples se procesan en el dispositivo, mientras que las complejas viajan a servidores cifrados donde, según Apple, ni la propia compañía puede ver el contenido.
- Siri AI ahora entiende texto, voz, imágenes y video simultáneamente, recuerda conversaciones anteriores y puede razonar sobre lo que la cámara captura, transformando el teléfono en un asistente que ve y piensa.
- Los usuarios de la Unión Europea quedarán excluidos de estas funciones al momento del lanzamiento, reflejando las crecientes fricciones regulatorias entre Apple y las autoridades europeas en materia de datos e inteligencia artificial.
- La beta para desarrolladores ya está disponible, pero con advertencias claras: inestabilidad, mayor consumo de batería e incompatibilidades con aplicaciones cotidianas; el lanzamiento público está previsto para septiembre.
En su conferencia anual de desarrolladores de junio de 2026, Apple presentó iOS 27 como una apuesta transformadora: Siri AI, el nuevo centro del sistema, funciona ahora con los modelos Gemini de Google, una alianza que rompe con la tradición de la compañía de construir sus capacidades de inteligencia artificial de forma independiente.
La arquitectura técnica busca conciliar potencia y privacidad. Las tareas cotidianas se ejecutan directamente en el dispositivo, mientras que las más complejas se envían a los servidores de Apple mediante un sistema llamado Private Cloud Compute, cifrado de tal manera que ni Apple puede acceder al contenido de las consultas. Es la apuesta de la empresa: ofrecer la capacidad de un gran modelo de lenguaje sin comprometer los datos del usuario.
Siri AI llega con su propia aplicación y puede invocarse desde el Dynamic Island. Su capacidad multimodal le permite procesar texto, voz, imágenes y video al mismo tiempo, recordar el hilo de conversaciones anteriores y razonar sobre lo que capta la cámara: desde extraer información nutricional de un menú hasta dividir una cuenta entre amigos. Las mejoras también alcanzan a Mail, Messages y Fotos, donde el sistema puede redactar mensajes en el tono del usuario, ampliar imágenes rellenando bordes de forma inteligente y eliminar objetos no deseados con resultados realistas.
Sin embargo, los usuarios de la Unión Europea no tendrán acceso a estas funciones en el lanzamiento, en medio de tensiones regulatorias no resueltas. Para quienes estén fuera de la UE y quieran anticiparse al lanzamiento público de septiembre, ya existe una beta para desarrolladores, aunque con las advertencias habituales: inestabilidad, mayor consumo de batería e incompatibilidades con aplicaciones como WhatsApp o plataformas bancarias.
Con iOS 27, Apple reconoce implícitamente que construirlo todo solo ya no es la estrategia más rápida ni la más eficaz. La pregunta que queda abierta es si esta alianza con Google, por más cifrada que sea la arquitectura, cambia de forma irreversible la promesa de privacidad que ha sido el sello distintivo de la compañía.
At its annual developer conference in June 2026, Apple introduced iOS 27, a system update built around a fundamental shift in how the company approaches artificial intelligence on mobile devices. The centerpiece is Siri AI, a reimagined version of the voice assistant that now draws its reasoning power from Google's Gemini models—a partnership that signals Apple's willingness to rely on external AI infrastructure rather than build everything alone.
The technical architecture behind this move is worth understanding. Apple designed a hybrid processing system that splits the work between your phone and remote servers. Simple, everyday tasks—the kind that don't require heavy computation—run directly on your device's processor, keeping things fast and keeping your data local. Anything more complex gets sent to Apple's servers through a system called Private Cloud Compute, which the company says is encrypted in a way that prevents even Apple itself from seeing what you're asking. This is the privacy argument: you get the power of a large language model without handing your requests to a third party.
Siri AI itself looks and behaves differently from its predecessor. It now has its own dedicated app on your iPhone, and it can be summoned from the Dynamic Island—that pill-shaped cutout at the top of newer iPhones. More importantly, it understands multiple types of input at once: text, voice, images, and video. This multimodal capability means the assistant can remember what you said in previous conversations and understand context. Point your camera at a restaurant menu, and Siri AI can extract nutritional information or suggest recipes. Photograph a bill with friends, and it can help you split the cost. This is visual intelligence—the phone learning to see and reason about what it sees.
The AI enhancements spread across Apple's built-in apps. In Mail and Messages, the system can draft entire messages from scratch, and it learns your writing style to match your usual tone. The Photos app gained spatial editing tools that let you expand an image by intelligently filling in the edges, correct perspective distortion after you've taken the shot, and remove unwanted objects while maintaining realistic backgrounds. These are features that previously required desktop software or manual work.
There is a significant geographic caveat. Apple confirmed that Siri AI and the advanced intelligence features will not be available in the European Union when iOS 27 launches. The company did not explain the restriction in detail, but it reflects ongoing regulatory tensions between Apple and EU authorities over data privacy and AI governance.
For users outside the EU who don't want to wait for the official September rollout, Apple has already released a developer beta. It's free to download, but it comes with the usual warnings: the software is unstable, it drains battery faster than finished versions, and it may break compatibility with everyday apps like WhatsApp or banking platforms. Anyone testing it should use a secondary device and back up their data first. The process involves registering as a developer on Apple's website, then navigating to Settings > General > Software Update and selecting the iOS 27 Developer Beta option. The system will then download and install it like any other update.
This release marks a turning point for Apple. For years, the company positioned itself as the privacy-first alternative to Google and Meta, building its own AI capabilities in-house. Now it's betting that partnering with Google's Gemini—while maintaining privacy through on-device processing and encrypted cloud compute—is the faster path to delivering genuinely useful AI features to hundreds of millions of users. Whether that bet pays off will become clear in the months ahead.
Citas Notables
Apple confirmed that Siri AI and advanced intelligence features will not be available in the European Union when iOS 27 launches— Apple (via announcement at WWDC 2026)
La Conversación del Hearth Otra perspectiva de la historia
Why did Apple choose to partner with Google for this? They've always built their own AI.
Speed, mostly. Building a competitive large language model takes years and billions of dollars. Google already has Gemini. By licensing it and wrapping it in their privacy architecture, Apple gets to ship powerful features this year instead of waiting three more years.
But doesn't that mean Google sees what users ask Siri?
Not according to Apple's design. The Private Cloud Compute system encrypts requests before they leave your phone. Apple says even they can't read what's being sent. Whether that's technically true and whether regulators believe it—that's a different question.
Why is the EU getting locked out?
The EU has been aggressive about AI regulation and data protection. Apple may have decided it's easier to launch without these features there than to fight over compliance. It's a temporary move, probably.
What's the actual difference between this Siri and the old one?
The old Siri was basically a voice-activated search engine. It could tell you the weather or set a timer. This one understands context, remembers conversations, and can reason about images and video. It can write for you, edit your photos intelligently, and help you solve problems, not just retrieve information.
Is it ready, or is the beta going to be a mess?
It will be a mess. Early betas always are. But that's the point—Apple's letting developers find the problems before September. If you're testing it, don't use it on your main phone.