No Siri AI in EU

This is so funny. Apple is unintentionally making a better product in the EU, no thanks to themselves.

Strong consumer and worker protections and less AI? I’m jealous of EU residents. Can we have that in America too?

So Europe gets Siri AI in 18+ months when TSA is complete.

That’s the overlap between what the EU wants and what Apple wants.

¯\_(ツ)_/¯

@Someone else Yes, but if you believe Apple’s public statements it won’t be ready in 18 months because they decided to stop working on it.

Apple is so ripe to be taken to the anvil and broken over it. The day is coming where the EU will simply say "no product or service from one company, may use exclusive means, or preferential access to interact with any other product or service from that same company, and may only connect via openly documented interfaces."

Apple is going to push until the final outcome is a blanket regulation so broad, so generic, that they're no longer allowed to tie their operating systems to their own devices.

There is not enough popcorn in the world for what's on the way.

@Someone: I really hope you are right, although I'm not nearly as optimistic. Imagine the bliss of running some flavor of Linux on my iPhone... And using it for whatever time it takes for the hardware to crumble to dust, instead of the few years that Apple deems it "supported"...

But I'm afraid it's just a wet SciFi dream.

@Michael, I don’t see Apple saying they’ve stopped working on anything.

In fact, just the opposite, as their press release is titled:
“ Due to DMA, Siri AI **delayed** in EU for iOS 27 and iPadOS 27” (emphasis mine)

“Delayed” implies that Apple plans for Siri AI to get to the EU at some point, no?

I DO see the EU commission saying things like
“What Apple is however not allowed to do, just like any other gatekeeper, is to close the market.
It is not for them to decide who gets to innovate in Europe.
And it's not for them to choose which AI tools our EU citizens get to use or not.”
(from the transcript on https://audiovisual.ec.europa.eu/en/media/video/I-290903)

Well, actually, it seems Apple *does* get to choose which AI tools EU citizens get to use or not, and Apple’s made their choice, which is to omit the EU from it’s Siri AI launch plans (until they can do so in a way that meets their privacy standards, which personally, I think is totally fine)

"Well, actually, it seems Apple *does* get to choose which AI tools EU citizens get to use or not"

They got to stop EU citizens from using Apple's LLM, which is fine; they're not required to provide access to it. This is like Microsoft being told they can't bundle IE with Windows and then being like, "Fine, no IE at all!" I mean, okay. That also solves the problem.

"until they can do so in a way that meets their privacy standards"

LOL.

"EU users will be able to access Siri AI on macOS 27 and visionOS 27." So your headline "No Siri AI in EU" isn't exactly accurate.

@Someone else The claim that Apple stopped working on it comes from Numerama, who interviewed Joswiak (SVP) directly. The press release doesn’t say either way. The “won’t ship until it meets Apple’s privacy standards” is the false narrative that the press release is trying to create, as I explained above.

Your link is broken. Here’s I think the same one:
https://www.numerama.com/tech/2271421-ils-ont-completement-ignore-nos-preoccupations-apple-explique-pourquoi-siri-ai-narrivera-pas-en-europe.html#:~:text=The%20result%20is,or%20even%20years.

I don’t see the false narrative you’re claiming. Apple said it’s doing what it’s doing for privacy reasons — what’s false about that?

Apple stands for privacy. That is their value proposition for their products, and Apple users get privacy whether they actively want it or not.

You may like it or not, but I value coherence. In Apple’s view, personal products such as phones should be privacy-first, and on that point they are coherent.

I agree that it is user-owned data. Apple’s position is: if you put your data into our products, we do not give it away, we do not access it, and we do not share it. Part of Apple’s cost is due to privacy and security, and security has a price.

At the very least, Apple does not make money by selling our data, unlike many others. It makes money by selling privacy-first products.

So when the EU says that Apple must share user data with any AI system provider, Apple cannot simply say yes. Free-market principles are at stake. A company should be able to create and sell the products it wants to make, at the price it wants to charge. If I want to sell a secure product, you cannot force me to make it insecure and change my entire business model.

Apple has clearly proposed TSA as an alternative: data would be shared only in a secure, user-controlled way. So why is TSA not ready? The problem is the size of the EU market. Developing and deploying TSA only for the EU costs money and time, and apparently Apple has decided that the EU market does not justify the effort.

Will Apple eventually do it and allow EU customers to take part in the AI revolution? I hope so, but I fear they will not. In my view, it is up to us to change the DMA into something more compatible with free-market principles.

Apple puts a high value on privacy and protecting user data. I know a lot of you don't believe this, but I think they go to great lengths to deliver on that. So I believe them when they say that no one, not even Apple, has access to identifiable information as a result of using the Siri AI on-device and private cloud services. It would be hard for them to say that they protect user privacy if they let any old model from ChatGClaudGrokemini have access to all of that.

"...And using it for whatever time it takes for the hardware to crumble to dust, instead of the few years that Apple deems it "supported..."

Fact: My Apple 12 mini (I refuse to upgrade the hardware if it means a larger screen size) was purchased in November 2020, over 5.5 years ago. My plan is to upgrade to iOS 27 next September, as it's supported. While not as long as my 2005 Dodge Neon (72,000 miles when sold after 16 years), it begs the question... how many here keep their "phone" (iPhone or other, including Android and Blackberry back in the day) for 5+ years?

I get a new iPhone every 2 years. My wife gets my "old" one, and my youngest daughter gets her old one, and I trade in that oldest one. So, collectively, we keep each iPhone about 6 years.

@DJ

While I agree that Apple does seem to take privacy more seriously than others, I would argue that they could review any AI that wanted to use the same frameworks Siri AI uses and reject any that do not meet the same privacy standard.
The EU wont complain if Apple can show that they are applying the same standard of data access to themselves and others, the problem is if Apple isn't doing that. Also, Apple hasn't yet built the framework that would actually do any of this. There does not exist (today) any framework that SiriAI uses that could be available for public consumption.

@Dave
I'm on an iPhone 13 mini that is now about 4 years old and still run iOS 18. I have no intention of upgrading to 26 or 27 and will eventually have to replace this phone. My phone before the 13 mini was an iPhone X that I had until it died at about 4 years old (the waterproofing failed and it was in my pocked when I went for a swim).
I am not really enamoured enough with any of the new AI stuff to bother with upgrading but I do like my tiny phone.

@Benjamin My understanding is that Siri AI will be labeled as "beta" even when the 27s are released in the fall, so I'm not surprised that they're not ready to "host" other AIs yet.

I'm not a fan of the EU's DMA stuff (in case anyone was wondering). Apple should tell them to pound sand.

@Someone else Thanks, I fixed my link. It was missing the final “l”. The explanation sounds false because Apple says:

1. They won’t ship it until they can make it private.
2. They know how to make it private.
3. They didn’t have that ready to go alongside Siri AI.
4. They’ve not even working to finish TSA.

The logical conclusion is that they are not actually trying to meet their professed standards. They care about privacy when it lets them be anticompetitive—they were hoping for an exemption—but then, e.g., record everything you type into the App Store and link it to your account, ignore privacy bug reports if you don’t run to the press, etc.

@DJ if Apple tells them to pound sand the EU should respond in kind, provide Apple a nice exit strategy for EU, and create a tech innovation fund for something like Murena or Sailfish to create an EU based alternative to Google.

@DJ I don’t see a consistent principle here. There are all sorts of ways that apps are already allowed to access private data, especially on Mac but even on iOS. You can opt into sharing photos, passwords, mail, calendars, Web pages you visit, location, etc. But when it’s in the context of AI it suddenly becomes too dangerous to even contemplate.

"So I believe them when they say that no one, not even Apple, has access to identifiable information as a result of using the Siri AI on-device and private cloud services."

I don't even understand what that means, because the demos they show involve sending identifiable information to Apple.

"Apple should tell them to pound sand."

Oh no. Imagine the infinite sadness of not having access to Siri AI.

@Michael,

“1. They won’t ship it until they can make it private.
2. They know how to make it private.
3. They didn’t have that ready to go alongside Siri AI.
4. They’ve not even working to finish TSA.

The logical conclusion is that they are not actually trying to meet their professed standards.”

#1, 2, 3: sounds right. That would logically lead to Apple’s actions… or non-action, as it were. Nothing unusual there.
#4: You don’t really know that except for a paraphrase from a French website translated into English, and which no one else has independently confirmed (so far).

On the other hand, Apple itself said in the title of its PR that it would take time, and not “never”. So logically…

@Someone else Apple’s PR says that there’s “no timeline” and bizarrely blames that on the EC’s “failure to acknowledge these risks.” What does that have to do with whether TSA is working or not? I don’t know exactly what’s going on, and this is not the first time that Apple told Numerama something that gave a different impression than what it told other people. But I think if Apple’s head of marketing gave an interview and the very same article got something of this importance wrong, either the article would be corrected or Apple would make a statement. And Joswiak would not have gone to Lellouche again if Apple were unhappy with his work last time. For whatever reason, this is what they want us to think.

@Dave: I'm currently still using my iPhone SE (1st generation).

@Michael, Logically, I’d wonder if the problem is Numerama, then, yeah?

Apple’s press release that you linked to is titled “Due to DMA, Siri AI **delayed** in EU for iOS 27 and iPadOS 27“

Apple has shareholders and they need to be truthful to them otherwise there will be lawsuits, and I believe they just learned this lesson recently WRT AI.

Re: “No timeline” — I have no timeline for my death, but I understand it’s coming someday (or so I hear).

@Michael Although I’m usually a fan of European regulation, in this case I find Apple’s reasoning pretty convincing.

Apple integrates Apple Intelligence as operating system service. The preferred execution environment is the local device. If necessary, the operating system decides whether to fall back to the cloud. Through Private Cloud Compute, Apple guarantees that even when data is processed in the cloud, no one other than the user has access to the processed data.

Apple can only provide this guarantee if it has complete control over the execution environment of the respective AI models.
For Gemini, they obviously have corresponding contracts with Google. That is likely to come at a cost.

Granting the same status to arbitrary models (in external cloud environments) would mean that the user completely loses control over access to their data. Due to its tight integration into the operating system, this is in no way comparable to an app.

One should also consider who is on the EU Commission’s side in this case… Every AI provider that would like to have unlimited access to the most private data of iOS users.

That doesn’t mean Apple hasn’t used similar arguments in the past to hinder the competition. But in this case, the reasoning seems sound to me.

Whether Apple will actually be able to run Apple Intelligence securely, however, is another matter entirely. Given how deeply integrated it is into the operating system, the notorious vulnerability of LLMs to attacks involving malicious input data strikes me as highly problematic.

Maybe it’s not such a bad thing that I have to wait and see how this plays out in the rest of the world before jumping in here in Germany. 😉

Welp, here’s corroboration from a second source that Apple’s not working on something that sounds a lot like TSA:

https://daringfireball.net/linked/2026/06/11/apple-dma-siri-ai#:~:text=what%20Apple%20is%20saying%20here%20is%20that%20they%E2%80%99re%20unwilling%20to%20create%20the%20open%2Daccess%2Dto%2Duser%2Ddata%20system%20that%20the%20EC%20is%20demanding
> Unless one side changes its mind and concedes its current position, Siri AI will never come to the EU, and what Apple is saying here is that they’re unwilling to create the open-access-to-user-data system that the EC is demanding.

So let’s see what happens in a few years — I’m guessing that the private off-the-record talk is bluffing (again, shareholder lawsuits) and that Apple is actually working on some privacy-guarding middleware between the spotlight semantic index and any other AI or app.

…Because that’s layer would be useful to/necessary for apps in the USA that say, want to control other apps or grab photos of mom but in a privacy-protected way.

Correction: “open-access-to-user-data system that the EC is demanding” (e.g. spotlight semantic index) *doesn’t* sound like a privacy-protecting TSA that we’re talking about.

So I think Numerama’s reporting that no work is being done by Apple is still uncorroborated.

Apple doesn’t want AI in EU with the same privacy privileges as SIRI AI, and Gruber is saying they won’t grant it.

Doesn’t mean there isn’t an eventual solution or that one (that takes 18+ months) isn’t in the works.

@Horst I don’t really see any reasoning there, just an axiom that Apple is more trustworthy and that their code does what they tell us it does. Some other company could provide models in a similar execution environment, and it might perhaps be faster or more secure in some way. In other areas where there’s competition, e.g. browsers, we find that Apple is slower to patch vulnerabilities. Or maybe another developer could do something really neat that works purely on-device and so is actually more private or functional than Apple’s solution.

@Dave

iPhone 3GS > 7 years
iPhone SE > 7 years
iPhone SE3 > 2 years so far.

Never replaced a battery, or a screen.

@MichaelTsai

Honestly, I think we're going to get to a place where software architecture will be as thoroughly regulated as building architecture, and that regulation will be to focus upon the integrative connection points. Look at the regulations for a staircase, or a doorway, and then imagine how much of a sword of Damocles the software world is toying with, by acting like obstinate children.

In macOS, Finder could theoretically be prohibited from having any privilege of access that third party file managers cannot have, etc.

Effectively, nothing user-facing would be allowed to be a "core system component". Third party replacements for System Settings, for example would be possible, and legally protected.

"Granting the same status to arbitrary models (in external cloud environments) would mean that the user completely loses control over access to their data"

This is exactly backward. In Apple's model, the user has no control over access to their data; Apple controls it. In the EU's model, the user regains control over access to their data and determines whom to grant access to it.

But as a company, Apple lists privacy of user data as one of its core values, and they tout that in a lot of their marketing. If Siri AI makes a mistake with that privacy, that's on Apple. If ChatGClaudGrokemini makes a mistake with that privacy, that's also on Apple, at least in the eyes of Joe User.

If you don't like how Apple does things, find another platform.

"Apple lists privacy of user data as one of its core values, and they tout that in a lot of their marketing"

Right, and if you buy into that, you have the choice of sticking with Siri AI.

"If ChatGClaudGrokemini makes a mistake with that privacy, that's also on Apple, at least in the eyes of Joe User."

No, it's not.

"If you don't like how Apple does things, find another platform."

There are two fucking platforms, and they're both shit. What exactly do you want me to do? Spend the next decade writing my own operating system?

Good. This shows to me that the DMA is working. As a gatekeeper you cannot benefit yourself and shut out competition. Not even for 18 months.

Apple also hugely disadvantages themselves with it. If people want AI on their phone and Samsung, Google, et al. do it in a way that complies with the DMA (which seems to be largely the case thus far), they extend their head start by even a larger amount than they had now.

"Apple designed a solution called Trusted System Agent — an intermediary that would allow virtual assistants to safely access the same features and capabilities as Siri AI for devices in the EU."

It wouldn't be beyond Apple if they asked 30% of every token bought that goes through the Trusted System Agent.

"the ability to act on that access autonomously without a user’s ongoing visibility and control"

They mean like the US can access data of EU citizens (through the CLOUD Act) without a user's ongoing visibility and control, by not making iPhone backups end-to-end encrypted by default? (And remember, your iMessage backups are not end-to-end encrypted because your friends don't have ADP enabled.)

---

I agree with many of the other messages: it seems like EU iOS gets better than US iOS by the year.

@DJ "Apple puts a high value on privacy and protecting user data. I know a lot of you don't believe this, but I think they go to great lengths to deliver on that. So I believe them when they say that no one, not even Apple, has access to identifiable information as a result of using the Siri AI on-device and private cloud services."

They don't have to because: 1. most people enable iPhone iCloud backups, since the flow is designed so that most people use it; 2. most people don't know that iCloud backups are not end-to-end encrypted by default; 3. most people do not have an idea what Advanced Data Protection is; and 4. even my iMessage message and I think also WhatsApp messages are stored without E2E because most people I communicate with don't have ADP enabled.

So, they willingly put data of EU citizens at risk of US law enforcement requesting it.

Apple's security/privacy policy is mostly oriented towards other companies not getting the data, a marketing thing, and a way of trying to shoot down regulations 'because privacy'.

@Daniël

"Apple designed a solution called Trusted System Agent — an intermediary that would allow virtual assistants to safely access the same features and capabilities as Siri AI for devices in the EU."

Yup, and if Apple had made Trusted System Agent the one and only way Siri AI accessed the device, they wouldn't be in this mess in the first place.

‘When the Chinese government wanted access to user data, Apple complied, but when the EU parliament voted for their citizens to be able to choose how to share their own data, that put Apple on a “war footing.”‘

I wish a journalist would ask Tim Cook this question. What a perfect way to phrase the obstinacy that Apple is showing with their behavior, here and for the past several years.

The best designs come from constraints. Apple seems to eschew that philosophy now in favor of some sort of might is right.

@Michael great write-up of the situation. Thanks (also to the many other people that you quoted) for consistently seeing through Apple's smoke screen when it comes to these opportunistic political games that Apple is playing.

> it begs the question... how many here keep their "phone" (iPhone or other, including Android and Blackberry back in the day) for 5+ years?

I do. My SE2 was 6 years old when I swapped it. My other AOSP thingy is 7 years old. I don't know what I would replace it with.

Can't be arsed to replace phones every few years. Not worth it.

> When the Chinese government wanted access to user data, Apple complied, but when the EU parliament voted for their citizens to be able to choose how to share their own data, that put Apple on a “war footing.”

Their PR behavior is what is most distateful from Apple whenever they fight the EUC. Indeed, that they exercise their options for recourse in front of a juridiction rather then just roll over as they do in China, since there is no recourse there, is the pride of liberal democracies. But for Apple to try and make their case to the public every time things don't go their way in the EU is both pointless, juridically speaking (there is no jury pool to taint), and exposes their hypocrisy when they don't expend any similar PR effort before complying, when it comes to China. This has to backfire on them, eventually.

@GioGio Except for the cases where Apple users don’t get privacy, and have no choice in the matter. Apple seems to care a lot more about privacy in cases where they can use it as a reason to thwart competitors.

I’m not sure free market principles are really at stake because we’re talking about a duopoly, not a free market. The DMA is a (perhaps clumsy) attempt to restore some market forces.

It seems like you agree with me that Apple could probably be shipping Siri AI if they had taken the time to build the TSA but that they chose instead to try to score PR points by implying that the EU wouldn’t accept that.