As Apple’s global The developer conference keynote concluded on Monday that market observers couldn’t help but notice the company’s share price has fallen, perhaps in comparison to Apple’s relatively low-key approach to AI than most competitors. Still, Apple Intelligence-based features and upgrades are abundant, and while some features are powered by the company’s privacy and security-centric cloud platform called Private Cloud Compute, many platforms run locally on Apple Intelligence-enabled devices.
Apple’s new message filtering feature automatically moves text from accounts and accounts you never interact with to the Unknown Sender folder. The feature automatically detects time-sensitive messages like login codes or food delivery updates and will still deliver them to your main inbox, but it also scans for messages that seem to be a scam and puts them in a separate spam folder. All of these sorting is done locally using Apple Intelligence. Similarly, the extended call filtering feature will automatically and locally answer untrusted calls, ask for details about the caller, and then transcribe the answer so that you can decide whether you want to answer the call. Even real-time translation adds real-time language translation to calls and messaging using local processing.
From a privacy perspective, local processing is the gold standard for AI capabilities. Data never leaves your device, which means there may be unconscious risks due to the journey through the cloud. And new features such as spam and “unknown senders” such as messages, call filtering for untrusted phone numbers, and real-time translation tools all seem to have designed a strategy to use privacy as a differentiation in already crowded AI fields.
In addition to being privacy-friendly, local processing has other benefits, such as allowing AI-based services to be offline and speeding up certain tasks because data is not necessary to be sent to the cloud, processed and sent back to the device. However, if AI capabilities will be widely available and accessible, most companies will limit them by trying to consider old low-end devices that many customers may not be able to use, which may not be able to handle local AI. However, Apple does not need inclusion because it produces both hardware and software and has imposed restrictions that Apple Intelligence can only run on recent device models.
Apple Intelligence also has other limitations, and the company also offers options for integration with some third-party generated AI services to expand functionality. For example, for OpenAI’s CHATGPT, the user must turn on the integration, and the Apple service will prompt the user to confirm each time the ChatGpt query is submitted. Additionally, users can choose to log in to the chatgpt account, in which case their queries will be subject to the normal policy of OpenAI, or they can use chatgpt without logging in.
Apple has invested heavily in developing private cloud computing to maintain strong security and privacy assurances for AI processing in the cloud. Other companies have even begun creating similar secure AI cloud solutions for products and services, with plans that specifically focus on privacy as a key feature. But the fact that Apple is still deploying local processing where possible may suggest that privacy is more than an intellectual priority for the company’s AI approach, which could be a business strategy.
