When Is Apple Intelligence Coming to Europe? It’s Complicated
Toggle Dark Mode
Not long after Apple showed off Apple Intelligence at its Worldwide Developers Conference (WWDC), it quickly clarified that the powerful new AI features would not be launching the European Union (EU) any time soon due to the “regulatory uncertainties” in that region.
Some wondered if this was a passive-aggressive move on the company’s part, resulting from the European Commission’s hardline stance on app distribution and some of Apple’s other policies in the EU under its new Digital Markets Act (DMA). However, it’s important to note that Apple didn’t say Apple Intelligence would never come to the EU, but merely that it needed some time to figure things out.
It’s not just Apple Intelligence that’s affected by the DMA. Apple also said that SharePlay screen-sharing and iPhone Mirroring to a Mac — both of which are expected to arrive in iOS 18 and macOS Sequoia this fall — would also be omitted from the EU, stating that “the interoperability requirements of the DMA could force us to compromise the integrity of our products in ways that risk user privacy and data security.”
The implied concern here is that European regulators could declare Apple anti-competitive by restricting developers from building apps to access these features.
Since Apple controls both ends of its screen mirroring technology, it can guarantee that everything is secure. Opening this up to third-party apps would be more complicated.
It’s not impossible, of course — Apple can create new APIs for this, just like it’s done for other iOS features — but it takes time to build and test these things. So, rather than delaying these features for the rest of the world while it retools iOS, iPadOS, and macOS to satisfy the EC, it’s chosen to omit those features from the more contentious market.
It’s a bit less clear where Apple Intelligence fits into the DMA, but it may be largely irrelevant for now. This fall, Apple’s AI features will launch solely in US English with a big, bold “beta” label slapped on. Apple Intelligence won’t likely be ready for prime time until early 2025, by which time Apple may have figured out how to navigate the vagaries of the DMA.
However, the DMA may not have been the only thing on the minds of Apple executives when they decided to delay the launch of Apple Intelligence. It seems the European Commission has yet another moving target that Apple will have to contend with: the European Union Artificial Intelligence Act.
Europe’s AI Act Creates More Uncertainty
The European Parliament approved the AI act in March, and now it’s slated to go into full effect on August 1, following its publication in the EU’s Official Journal.
While the EU AI Act is mostly about regulating “high-risk” uses of artificial intelligence, such as biometrics and applications in law enforcement and critical infrastructure, it also demands transparency from developers of AI tools like OpenAI and Google.
The requirements of the AI Act will come into force in a phased approach. Specific “unacceptable risk” cases will be banned immediately, which isn’t a bad thing. For example, AI can’t be used for social credit scoring or creating facial recognition databases from random internet scraping or compiling CCTV footage.
By next spring, all AI developers — which will include Apple if it launches Apple Intelligence in the EU — will need to agree to codes of practice that haven’t been written yet.
The European Union’s AI Office plans to call for stakeholders to express an interest in helping to draft these codes, and Apple could undoubtedly throw its hat into the ring. However, it’s still facing the challenge of launching Apple Intelligence in a market where it may be forced to either agree to at-yet-undefined terms or withdraw its AI features from the EU entirely.
In the face of that sizeable question mark, it’s probably much wiser to avoid introducing these features in the first place. “Regulatory uncertainties” indeed.
Finally, by next summer, the EU AI Act will require General-purpose AI (GPAI) systems, which would include ChatGPT, Gemini, and Apple Intelligence, to meet specific transparency requirements. This includes providing detailed summaries of how these models are trained and the content used to train them.
The Act also requires companies to ensure that generative AI media is clearly labeled in accordance with whatever standards the EU has yet to specify and that controls are in place to prevent copyrighted material from being captured by their training models without express authorization from rights holders.
Those last goals are laudable, and Apple can and should comply with them, but it won’t meet those requirements unless it changes its ways. Many creative artists are already concerned about where Apple is getting its data, and at this point, Apple is only allowing sites to opt out of having their data used for AI training. If a site hasn’t done that, Apple’s AI models assume the data is fair game.
The EU’s AI Act will require sites to explicitly opt in before their data can be used for this purpose. In a dispute, the onus will be on GPAI companies like Apple, Google, or OpenAI for not performing their due diligence before using copyrighted material.
