I don't think Google would want to foot the cost of the processing, and I don't think Apple would want to give Google all of that data. Apple is also big on owning their core technology, so paying for a plug-in AI support that they control, until they can eventually spin-up their own to be good enough, seems like the play.Privacy also matters. I think a lot of people who use Apple products do so because they tout privacy so much. If it turns out they're handing Siri data to Google on a silver platter, it's going to erode that reputation.
![]()
Privacy
Apple products, including Apple Intelligence, are designed to protect your privacy, because privacy is a fundamental human right.www.apple.com
If this turns out to be true, I personally would have less of a problem with it, I think (pending any new information about the privacy of said data).
100% agree. I would love to see it totally provider-agnostic so each user can pick and choose their provider based on what they value (privacy, environmental footprint, ethical considerations, respect for intellectual property, etc etc).Letting users pick their model like they can with search engines should still be the North Star.
Locking into a singular model long term will be bad for Apple and end users.
Hopefully Apple at least secured data privacy protections or else Siri will get turned off.
Let's be real: Apple does not really care about tracking/privacy and has only said they do for marketing reasons. Apple is tracking you for their own ad spaces, does not pull apps from large companies tracking you against the TOS of the App Store and/or your preferences and the company only makes decisions that benefits the shareholders + bottom line. Apple does the bare minimum in an industry where others don't do the bare minimum, which elevates them from a marketing standpoint.I don't like this at all. Google is a tracking hound, and Apple should know better.
In AI's case, the larger the better? just wondering why. I would prefer a more accurate and compact version, and smarter ofc.as Gemini's large language model is significantly larger than Apple's own model.
That’s a shame
I’ll be disabling Siri and retiring my old HomePod in that case 😕
Embarrassing that a company as large as Apple couldn't develop their own competitive SOTA LLM from scratch.
Gemini is really fine to use but that is thanks they sucked whole internet for free. This will be monopoly.This is exactly what Apple and Google’s strategy is here — to maintain the status quo and block any potential AI companies from building a competing ecosystem by teaming up.
Gemini is very open minded and you can chat with it on very sophisticated levels. I am afraid Apple will ruin this 😂Training a SOTA LLM requires 6m and 100M of compute (both just order of magnitudes, not exact figures). And a capable team. And training data.
Apple has plenty of capital to do this, but data might be a hard to fix gap:
- You have Apple TV original content; so you have access to a high quality dataset, but it is tiny
- You can license Reddit; so you have access to up-to-date large dataset
- You cannot license YouTube
I believe Apple to be dramatically more conservative in utilizing "freely available" training data than xAI, OpenAI, Meta, Anthropic, and likely Google. Would Apple dare to use:
- Any of their user data — I do not think so
- Apple Music — I do not think so
- Podcasts — Maybe... This is a treasure trove
- YouTube — I do not think so
- Reddit — Definitely, but would pay for it
- Z-Library —I do not think so
Did some of the others use many of these without right to do so — Absolutely.
Also, Apple is very risk averse on model output. They will not risk creating a MechaHitler, generating CSAM, or even generating Mickey Mouse picture. Others can move fast due to willingness to take risks.
With all this in mind — Apple likely can train a good quality multimodal LLM when they so choose, but it will likely be lagging behind the frontier model capabilities due to lacking training data access.
Inviting Google into iPhone is like inviting a vampire into your home!
In a statement shared with CNBC today, Apple confirmed that Google Gemini will power the next-generation version of Siri that is slated to launch later this year.
![]()
"After careful evaluation, we determined that Google's technology provides the most capable foundation for Apple Foundation Models and we're excited about the innovative new experiences it will unlock for our users," the statement said.
The report explicitly mentioned that Google Gemini will power Siri, as was expected. Apple's decision to lean on Google's artificial intelligence technology should result in the revamped Siri being more capable and advanced than it otherwise would have been, as Gemini's large language model is significantly larger than Apple's own model.
The next-generation version of Siri is expected to be introduced with iOS 26.4, which will likely be released to the general public in March or April.
Apple first announced the personalized Siri features during its WWDC 2024 keynote, but last year it announced that they were delayed. The new capabilities will include better understanding of a user's personal context, on-screen awareness, and deeper per-app controls. For example, Apple showed an iPhone user asking Siri about their mother's flight and lunch reservation plans based on info from the Mail and Messages apps.
Update: Apple and Google today announced that Gemini will help power not only a more personalized version of Siri, but a range of future Apple Intelligence features.
Article Link: Apple Confirms Google Gemini Will Power Next-Generation Siri This Year
I don't understand people sometimes. There is no using Gemini without Google getting some data in return. Unless Apple is cutting a pretty massive check for a version that runs exclusively on Apple private cloud compute.If no useful data is going to (presumably Apple's) Gemini servers, then how is Gemini going to be trained?
I'm pretty confident Gemini won't be on-device. Apple doesn't permit 3rd party's system level access for privacy protections. So I'm assuming Gemini will replace ChatGPT for cloud compute since it wouldn't make sense for them to contract both.Interesting question. I'm not so sure. If the model on device that is powered by Gemini cannot answer the question, it will still need to punt to the cloud for the answer. Does that then use actual, real Gemini or go back to ChatGPT like it does now? Apple pays nothing for their OpenAI access, so there is no advantage to removing it unless being all-Gemini, all-the-time provides a more consistent experience between things being done on device and things done in the cloud.