Here’s how Apple Intelligence protects privacy compared to the competition

Artificial Intelligence has been the craze in the industry for the last couple of years. Large language models are extremely capable, but they always come with one compromise: privacy. You’re still using the model hosted in cloud and all your conversations are stored somewhere.

Apple has always placed privacy at the heart of its products. In the months leading up the launch of Apple’s AI, many people wondered how Apple would handle it. Today, we’ll explore how.

On-device-models

It’s a bit obvious that Apple Intelligence is powered by models on the device. This is why there is such a strict requirement for the device. Apple Intelligence is only available if you own an iPhone 15 Pro, iPhone 16 or iPhone 16 Plus. All M1 devices are supported on the iPad and Mac.

This device requirement isn’t just arbitrary, it’s because running large languages models on a device requires a great deal of power and memory. Apple has only recently begun shipping the iPhone 15 Pro with 8GB of unified RAM.

Requests made on the device model don’t leave the phone. Apple Intelligence requests may not all run on the device. Apple has a private cloud compute system for heavy requests, as I’ll explain later. Apple Intelligence features such as Genmoji and Notification summaries rely heavily on the chipset of your device. This keeps your personal information and requests local.

Apple’s also expanded the capabilities of on device models this year at WWDC25, allowing developers to implement features using Apple’s on device models, otherwise known as Apple Foundational Models. This’ll potentially deter developers from using providers like OpenAI or Google Gemini for passing your private user data to, though there is the major fact that Apple Intelligence is only available on a small number of devices.

Private Cloud Compute

Apple has a cloud-based solution to handle Apple Intelligence requests. It wasn’t used much in iOS 18, but iOS 26 has seen it being used more. You can now use Siri Shortcuts for Apple models including cloud models. Apple has an incredibly comprehensive

Blog post about how Private Cloud Compute protects your user information. You can verify that the system was designed to not store information. Apple also wanted to release software images of Private Cloud Compute to allow independent researchers to verify Apple’s claims. Learn more about Apple’s Private Cloud Compute in our blog post. It’s an interesting system and is definitely industry-leading in terms of server-based AI security.

ChatGPT Integration

Apple has a special agreement [with OpenAI]to ensure that Apple data is not retained by OpenAI and that Apple requests are not used in training data for models. ChatGPT never sends requests to users without their permission.