Apple says the Digital Markets Act forced its hand. Brussels says Cupertino asked to be let off the hook entirely. Either way, 450 million Europeans will not get Siri AI on their iPhones this year, and the fight over who controls the assistant in your pocket has only just begun.

By Blumefield | June 10, 2026

A Launch With a Hole in the Map

Twenty-four hours after Tim Cook closed his final WWDC keynote with the biggest Siri overhaul in the assistant's history, Apple confirmed the launch will skip an entire continent. Siri AI, the rebuilt assistant at the heart of iOS 27, will not ship on iPhone or iPad in any of the 27 EU member states when the software arrives this autumn. In a statement published on Apple's newsroom, the company blamed the Digital Markets Act, saying months of talks with regulators produced no acceptable path to compliance.

The carve-out is unusually messy. EU users will get Siri AI on macOS 27 and visionOS 27, but not on iPhone, iPad or Apple Watch, since the watch version requires a paired iPhone running the new assistant. Developers based in the EU cannot even test Siri AI features for their own apps. The result is a two-tier Apple ecosystem inside a single market that prides itself on regulatory coherence, arriving at the precise moment Apple needs to prove it can compete in consumer AI at all.

The stakes go well beyond one product cycle. Siri AI is Apple's answer to two years of broken assistant promises, a $250 million settlement over advertised features that never shipped, and a reported $1 billion a year licensing deal that puts a custom Google Gemini model at the assistant's core. Europe is roughly a quarter of Apple's revenue. Shipping the company's flagship AI feature everywhere except its second-largest market is not a rounding error. It is a strategic wound, and both sides insist the other inflicted it.

What Brussels Actually Demanded

The Digital Markets Act requires designated gatekeepers, Apple among them, to make core platform functions interoperable with third parties. Applied to the new assistant, the European Commission's interoperability rules mean that whatever deep hooks Siri AI gets into messages, files, purchases and apps must also be available to rival assistants, whether from Google, OpenAI, Anthropic or a European challenger.

Apple's characterisation of that obligation is blunt. The company says regulators' reading of the law would force it to give any AI system "nearly unlimited access" to a user's device, including the ability to read and send messages, make purchases, access files and act across apps autonomously. Apple points to security research showing AI agents can be hijacked to exfiltrate passwords and photos or silently alter accounts, and argues those risks are growing as assistants gain capability.

The Commission tells a very different story. Spokesperson Thomas Regnier said on Tuesday that the decision not to ship "is Apple's and Apple's only," that the company "was simply unable to develop interoperability solutions" meeting EU privacy and security standards, and that instead of finding a compliant design Apple requested a blanket exemption from its obligations. That, Regnier said, "is not an option." The gap between the two accounts is not a nuance. One side describes a safety dispute, the other describes a refusal to follow the law.

The Trusted System Agent Gambit

Apple's proposed compromise deserves more attention than it has received. The company designed an intermediary layer it calls Trusted System Agent, which would let third-party assistants reach the same features and capabilities as Siri AI through a mediated, permissioned channel rather than raw device access. Apple says it offered to launch Siri AI in the EU immediately and phase in the new architecture over 18 months. The Commission, by Apple's telling, rejected every proposal on the table.

Viewed charitably, Trusted System Agent is a serious attempt to reconcile agentic AI with platform security, an API gatekeeper for the gatekeeper. Viewed sceptically, it is the App Store playbook transposed to AI: Apple decides what rivals can do, when, and through which toll booth, while keeping privileged access for its own assistant. Brussels has spent five years legislating against exactly that pattern, which explains why an 18-month phased rollout controlled by Apple landed badly with officials who already fined the company €500 million under the same law.

There is an irony underneath the privacy argument that neither side wants to dwell on. The Siri AI that Apple describes as private by design runs on a 1.2 trillion parameter Gemini model licensed from Google, albeit hosted on Apple's own Private Cloud Compute infrastructure. Apple is simultaneously arguing that third-party AI access to user data is too dangerous to permit, and that its own arrangement with the largest advertising company on earth is fine because Apple controls the plumbing. Both claims may be true. They are still awkward neighbours.

A Collision Months in the Making

This standoff did not appear from nowhere. Apple delayed Apple Intelligence features in the EU in 2024 over the same interoperability concerns, relented later, and has been escalating its public criticism of the DMA ever since. The difference now is timing and scale. As Blumefield has reported, the EU is heading into its largest enforcement moment yet, with the bulk of the AI Act applying from August 2 and fines reaching 7 percent of global turnover, while Brussels simultaneously reviews its digital rulebook under pressure to simplify it.

That context cuts both ways. European officials cannot afford to look like they blink when the world's most valuable company plays hardball, particularly with Washington framing EU tech regulation as a trade grievance. Apple, for its part, has watched the assistant wars accelerate around it. The new Siri arrives years behind ChatGPT and Gemini as a conversational product, and Apple's own WWDC announcement that users can choose ChatGPT, Gemini or Claude as their Apple Intelligence model shows the company already conceding that assistant choice is coming. The dispute is really about whether that choice arrives on Apple's schedule or on Brussels' terms.

There is also a quieter commercial logic. Withholding a marquee feature from European consumers creates a constituency for Apple's position. Every EU iPhone owner who wonders why their device cannot do what an American one can becomes an argument against the DMA, made at no cost to Apple's compliance record. Regulators understand this dynamic perfectly well, which is why the Commission moved within a day to pin the decision on Cupertino.

What Happens Now

The most likely path is the familiar one: months of technical dialogue, a revised Apple proposal with shorter timelines and less unilateral control, and a delayed EU launch sometime in 2027. Apple says it has no timeline but wants to bring the features to Europe. The Commission says the door to a compliant solution remains open. Neither side benefits from a permanent freeze, and the 2024 Apple Intelligence delay ended with a deal.

The harder question is what this episode signals for agentic AI everywhere. The DMA was written for app stores and browsers, not for assistants that can read messages and move money. Europe is now stress-testing whether interoperability mandates can coexist with AI agents that are themselves a security attack surface, and the answer will shape how Google, Samsung and every Android manufacturer approach assistant integration in the bloc. Expect rival labs to push for access rights under the precedent, whatever it turns out to be.

For Apple, the calculus is starker. The company has bet its AI credibility on a partner's model, a privacy architecture, and a fall release that now excludes a quarter of its revenue base. As Blumefield wrote in its WWDC coverage, Siri AI was supposed to be the moment Apple stopped apologising for its assistant. In 27 countries, the apology tour has simply changed venue, from the stage at Apple Park to the press room in Brussels.