An Actual AiPhone
By my count, this is my fifth headline trying to make 'AiPhone' happen. I should probably stop. But tonight is not the night, damnit. Because there is once again hope. Hope that Apple has finally taken their collective head out of the sand. That they're going to rise to meet the moment that is AI.
There have been signs. After the complete and utter debacle of Apple's first attempt to launch the AiPhone in 2024, the embarrassment stretched into 2025 with almost nothing to show for it. Heads finally rolled, but that still wasn't enough because it still wasn't clear that even with new people in place, that Apple would be on the right path with AI. After all, many of the key players who had spent the past several years downplaying the importance of LLMs, or of NVIDIA chips, or belittling chatbots were still in place.
Well, that finally changed a couple months ago and look at that: it now feels like Apple is being honest with themselves at least. Siri wasn't just a few tweaks away, Siri needed a lobotomy. And after a new brain bakeoff, they got one in the form of Gemini. Finally, a world-class back-end to power Siri. But she was still sort of stuck in 2016 mode. Basically a smart speaker version of AI. Well now Siri may be on the verge of getting a new front-end to match the new back-end too, it seems.
As Mark Gurman reports for Bloomberg:
Apple Inc. plans to revamp Siri later this year by turning the digital assistant into the company’s first artificial intelligence chatbot, thrusting the iPhone maker into a generative AI race dominated by OpenAI and Google.
The chatbot — code-named Campos — will be embedded deeply into the iPhone, iPad and Mac operating systems and replace the current Siri interface, according to people familiar with the plan. Users will be able to summon the new service the same way they open Siri now, by speaking the “Siri” command or holding down the side button on their iPhone or iPad.
The keyword there, of course, is "chatbot". On the surface, that may not sound much different from current Siri, but again, instead of being the sort of one-off voice assistant, Siri should now be a full-on, LLM-powered AI.
The difference should be even more apparent because it sounds like we're still first going to get "old" Siri just with that new Gemini-powered brain in iOS 26.4 this Spring, before we get new Siri in iOS 27 this Fall. The first version will likely just be competent (finally), while the second version should actually be conversational. An actual AI product like we're used to from ChatGPT or yes, Gemini itself.
The previously promised, non-chatbot update to Siri — retaining the current interface — is planned for iOS 26.4, due in the coming months. The idea behind that upgrade is to add features unveiled in 2024, including the ability to analyze on-screen content and tap into personal data. It also will be better at searching the web.
The chatbot capabilities will come later in the year, according to the people, who asked not to be identified because the plans are private. The company aims to unveil that technology in June at its Worldwide Developers Conference and release it in September.
Campos, which will have both voice- and typing-based modes, will be the primary new addition to Apple’s upcoming operating systems. The company is integrating it into iOS 27 and iPadOS 27, both code-named Rave, as well as macOS 27, internally known as Fizz.
Gurman notes that this change may be the biggest revamp of iOS 27, after last year's big "Liquid Glass" UI changes. So it will be especially important to get this right. To take the iPhone and make it the AiPhone.
Internally, Apple is testing the chatbot technology as a standalone Siri app, similar to the ChatGPT and Gemini options available in the App Store. The company doesn’t plan to offer that version to customers, though. Instead, it will integrate the software across its operating systems, like the Siri of today.
Interesting that Siri still won't have an app in this new iOS 27 build. I'm not sure this isn't a mistake simply because that's how people – iPhone users in particular – know to interact with AI chatbots right now. Sure, you still want Siri running underneath, always on, as it were. But it would probably be helpful for users to have a hub – a Siri app – to load up. We'll see, I guess!
Like ChatGPT and Google Gemini, Apple’s chatbot will allow users to search the web for information, create content, generate images, summarize information and analyze uploaded files. It also will draw on personal data to complete tasks, being able to more easily locate specific files, songs, calendar events and text messages.
Unlike third-party chatbots running on Apple devices, the planned offering is designed to analyze open windows and on-screen content in order to take actions and suggest commands. It will also be able to control device features and settings, allowing it to make phone calls, set timers and launch the camera.
Yes, this all sounds great – if it actually works this time. It sounds like... the AiPhone.
One issue under discussion is how much the chatbot will be allowed to remember about its users. ChatGPT and other conversational AI tools can retain an extensive memory of past interactions, allowing them to draw on conversations and personal details when fulfilling requests. Apple is considering sharply limiting this capability in the interest of privacy.
Uh oh, this may be Apple falling back to their old ways of thinking again. We all understand why – and I've written about the potential AI memory "problems" in the past – but this may be another situation where the market has spoken. If all the other popular AI services have great memory features that remember a lot about you but Apple limits this, it could hurt the product.
Beyond that, Gurman goes into the weeds about the differences in foundation model terminology for the first new version of Siri we're going to see – seemingly in just a matter of weeks – versus the one due this Fall. One big thing of note:
In a potential policy shift for Apple, the two partners are discussing hosting the chatbot directly on Google servers running powerful chips known as TPUs, or tensor processing units. The more immediate Siri update, in contrast, will operate on Apple’s own Private Cloud Compute servers, which rely on high-end Mac chips for processing.
It's fairly well known that Apple already uses Google Cloud for some of their infrastructure – and TPUs for much of their AI work (as they still have that pesky NVIDIA aversion), so this shouldn't be too shocking. Still. I'd be surprised if Apple doesn't tout this as running some variation of Private Cloud Compute in Google Cloud.
Then again, Apple is clearly building this new Siri to be more extensible. The early talking points will be so that they can operate in China, which will obviously not allow Google-powered models powering Chinese iPhones. But eventually, this will let Apple swap out Gemini for other AI brains as well – ideally, one day, their own. That will be the true AiPhone, I suppose. For now, this will have to do.
One more thing: is there a world in which Apple uses the big, new AI upgrade in iOS 27 to also rebrand Siri? It sounds like that's not in the cards right now, but it obviously also might not be the worst idea given Siri's past stumbles...
M.G. Siegler
M.G. Siegler
M.G. Siegler
M.G. Siegler