Apple may be preparing to open its artificial intelligence layer to outside providers, allowing users to swap in their preferred large language model rather than relying solely on the company's own systems. The Verge reported the development, citing Bloomberg's Mark Gurman, who writes that Apple is planning to let third-party chatbots power Apple Intelligence features across iOS 27, iPadOS 27, and macOS 27.
To understand why this is significant, it helps to recall how Apple arrived at this point. When the company unveiled Apple Intelligence at its developer conference in 2024, it positioned the suite as something meaningfully different from the AI race happening elsewhere in the industry. The pitch was privacy-first, on-device where possible, and tightly controlled. Apple brokered a high-profile arrangement with OpenAI to bring ChatGPT into Siri as an optional extension, but the framing was careful: Apple was the gatekeeper, users would be asked before any query left Apple's infrastructure, and the OpenAI integration was presented as a curated partnership rather than an open door. The underlying message was that Apple Intelligence meant Apple's judgment about which AI you should be using.
What Gurman is now describing would represent a meaningful shift in that philosophy. Moving from a single vetted partner to a broader menu of selectable models is not a trivial architectural decision. It implies Apple is building something closer to an AI abstraction layer inside its operating systems, a kind of switching mechanism that routes requests to whichever model a user has designated. That is a more complex engineering commitment, and it is also a more exposed one, since Apple would be extending trust to a wider range of external systems that it does not fully control.
The competitive pressure behind this move is not hard to identify. Google has been aggressive about weaving its Gemini models into Android at a deep level, and Samsung has made its own AI partnerships a centerpiece of its device marketing. Meanwhile, the landscape of capable frontier models has expanded considerably. Anthropic, Meta, Google, and several smaller players now offer models that a meaningful segment of users might genuinely prefer over whatever Apple or OpenAI provide. Locking those users into a single provider increasingly looks less like a premium experience and more like a constraint.
There is also a regulatory dimension that almost certainly factors into Apple's thinking. Antitrust scrutiny of platform gatekeeping has intensified in the European Union and, to varying degrees, elsewhere. Apple's exclusive AI arrangements could attract the kind of attention that its App Store policies already have. Offering user choice preemptively is a posture the company has employed before, and opening the AI layer could serve as a useful argument that Apple is not unduly favoring its own services or any single commercial partner.
For users, the likely consequence is a more genuinely customizable AI experience, at least for those who follow the space closely enough to have a preference. Most mainstream users will probably leave whatever default Apple ships, but the option matters symbolically and will matter practically to power users, developers, and enterprise customers who have built workflows around specific models. For third-party AI providers, the prospect of system-wide integration with Apple's installed base is an enormous distribution opportunity, and the competition to become a listed option or, better still, a popular default will be intense.
For Apple itself, the risks are real even if the strategic logic is sound. Quality control across a range of external models is harder than quality control over one. If a third-party model produces harmful, embarrassing, or simply poor output while operating inside Apple Intelligence, the reputational damage lands on Apple's brand as much as on the model provider. Apple will need to think carefully about what guardrails, certification requirements, or liability frameworks govern participation, and those decisions will shape which players actually make it onto the list.
The arrangement with OpenAI also bears watching. That partnership was presented as the premium external option, and it presumably came with commercial terms reflecting that position. A world in which OpenAI is one choice among many is a different commercial reality than one in which it is the designated outside voice of Apple Intelligence.
The features described by Gurman are tied to software releases that are expected to be previewed at Apple's annual developer conference, likely in the summer, with public releases following in the autumn. Between now and then, the key things to watch are which AI providers Apple announces as supported, what Apple says publicly about how third-party models are vetted, and whether the company frames this as a user empowerment story, a competitive response, or something else entirely. How Apple chooses to tell this story will reveal as much about its strategic intentions as the feature itself.
