Apple is opening up access to the same generative models that power its Apple Intelligence system, giving third-party developers the ability to bring new AI features into their apps. The announcement came during the company’s Worldwide Developers Conference (WWDC 2024), where Apple revealed its new Foundation Models framework. With it, developers can integrate AI-driven tools into their software for iOS 18, iPadOS 18, and macOS Sequoia. That means apps can handle tasks like summarizing text, generating images, or suggesting replies without the need to build complex AI systems from the ground up.
Key Takeaways
- Developers now have access to Apple’s own generative AI models within their apps.
- The framework supports on-device text generation and image creation.
- Privacy remains central, with on-device processing and Private Cloud Compute.
- The tools are part of iOS 18, iPadOS 18, and macOS Sequoia.
What the New Framework Offers
The Foundation Models framework is essentially a set of tools that lets developers tap directly into Apple’s generative AI capabilities. Rather than spending vast resources to train their own models, developers can simply use an Application Programming Interface (API) to call on Apple’s models for specific functions.
Take a note-taking app, for example. With just one added feature, it could summarize long passages of text almost instantly. A design app might let users create custom images by typing a short description. Messaging apps could become smarter, offering more context-aware reply suggestions that feel natural rather than robotic.
Apple has designed the system to operate in two layers, balancing functionality with privacy. For lighter tasks, the AI runs directly on the device itself. This approach is fast, efficient, and private, since no information leaves the iPhone, iPad, or Mac.
Privacy and More Complex Tasks
When a task demands more power than a device alone can handle, Apple Intelligence shifts to what it calls Private Cloud Compute. In these cases, the request is securely sent to Apple servers built on Apple silicon. According to the company, the data is never stored and is used solely to complete the request. Apple also emphasizes that its servers are designed to be verifiable, allowing independent experts to inspect the code that runs on them.
For developers, this setup removes the burden of managing user data themselves. It hands the responsibility for privacy and security over to Apple’s infrastructure. That makes the offering stand apart from other AI platforms, many of which process and retain user data on external servers. By providing these ready-made models, Apple is essentially lowering the barrier for developers to experiment with advanced AI, while at the same time reinforcing its familiar stance on privacy.
Frequently Asked Questions (FAQs)
Q. What is Apple’s Foundation Models framework?
A. It is a set of tools and APIs that allows app developers to use Apple’s built-in generative AI models to create features like text summarization and image generation within their own applications.
Q. Is it free for developers to use?
A. Yes, the framework is part of the standard software development kits (SDKs) for iOS 18, iPadOS 18, and macOS Sequoia, and there is no extra charge for developers to use these features.
Q. Which devices will support Apple Intelligence features?
A. Apple Intelligence requires the processing power of modern Apple silicon. It will be available on the iPhone 15 Pro and iPhone 15 Pro Max, as well as on iPad and Mac models with an M1 chip or later.
Q. Is my data safe when using these new AI features?
A. Apple has designed the system with privacy as a central point. Many tasks are processed entirely on your device. For more complex tasks, data is sent to Apple’s Private Cloud Compute servers, but Apple says the information is not stored or made accessible to them.
Q. Can developers use other AI models like those from OpenAI?
A. Yes. Apple also announced an integration with OpenAI’s ChatGPT. This means users can choose to send certain requests to ChatGPT for its world knowledge, but this is separate from the Foundation Models framework, which specifically uses Apple’s own models.
