Reactions to Apple Intelligence, which Apple unveiled at WWDC 2024, have ranged from curious to positive to underwhelmed, but whatever your views on the technology itself, a big talking point has been Apple’s emphasis on privacy, in contrast to some companies that have been offering generative AI products for some time.
Apple is putting privacy front and center with its AI offering and has been keen to talk about how Apple Intelligence – which will be integrated across iOS 18, iPadOS 18, and macOS Sequoia – would differ from its competitors by adopting a fresh approach to handling user information.
Craig Federighi, Apple’s Senior Vice President of Software Engineering, and the main presenter of the WWDC keynote, has been sharing more details about Apple Intelligence, and the company’s privacy-first approach.
Speaking to Fast Company, Federighi explained more about Apple’s overall AI ambitions, confirming that Apple is in agreement with other big tech companies that generative AI is the next big thing – as big a thing as the internet or microprocessors were when they first came about – and that we’re at the beginning of generative AI’s evolution.
Apple's commitment to AI privacy
Federighi told Fast Company that Apple is aiming to “establish an entirely different bar” to other AI services and products when it comes to privacy. He reinforced the messaging in the WWDC keynote that the personal aspect of Apple Intelligence is foundational to it and that users’ information will be under their control. He also reiterated that Apple wouldn’t be able to access your information, even while its data centers are processing it.
The practical measures that Apple is taking to achieve this begin with its lineup of Apple M-series processors, which it claims will be able to run and process many AI tasks on-device, meaning your data won’t have to leave your system. For times when that local processing power is insufficient, the task at hand will be sent to dedicated custom-built Apple servers utilizing Private Cloud Compute (PCC), offering far more grunt for requests that need it – while being more secure than other cloud products in the same vein, Apple claims.
This will mean that your device will only send the minimum information required to process your requests, and Apple claims that its servers are designed in such a way that it’s impossible for them to store your data. This is apparently because after your request is processed and returned to your device, the information is ‘cryptographically destroyed’, and is never seen by anyone at Apple.
Apple has published a more in-depth security research blog post going into more detail about PCC, which, as noted at WWDC 2024, is a system available to independent security researchers, who can access Apple Intelligence servers in order to verify Apple’s privacy and security claims around PCC.
Apple wants AI to feel like a natural, almost unnoticeable part of its software, and the tech giant is clearly keen to win the trust of those who use its products and to differentiate its take on AI compared with that of rivals.
More about ChatGPT and Apple Intelligence in China
Federighi also talks about Apple’s new partnership with OpenAI and the integration of ChatGPT into its operating systems. This is being done to give users access to industry-standard advanced models while reassuring users that ChatGPT isn’t what powers Apple Intelligence; the latter is exclusively driven by Apple’s own large language models (LLMs), which are totally distinct on Apple’s platforms, but you will be able to enlist ChatGPT for more complex requests.
ChatGPT is only ever invoked at the user’s request and with their permission, and before any requests are sent to ChatGPT you’ll have to confirm that you want to do this explicitly. Apple teamed up with OpenAI to give users this option because, according to Federighi, GPT-4o is “currently the best LLM out there for broad world knowledge.”
Apple is also considering expanding this concept to include other LLM makers in the future so that you might be able to choose from a variety of LLMs for your more demanding requests.
Federighi also talked about its plans for Apple Intelligence in China – the company’s second biggest market – and how the company is working to comply with regulations in the country to bring its most cutting-edge capabilities to all customers. This process is underway, but may take a while, as Federighi observed: “We don’t have timing to announce right now, but it’s certainly something we want to do.”
We’ll have to see how Apple Intelligence performs in practice, and if Apple’s privacy-first approach pays off. Apple has a strong track record when it comes to designing products and services that integrate so seamlessly that they become a part of our everyday lives, and it might very well be on track to continue building that reputation with Apple Intelligence.