Apple and OpenAI sold their partnership as a clean answer to the iPhone maker's AI problem. Two years later, it looks more like a warning about platform dependency.
The Apple-OpenAI alliance is no longer just a product story. It is becoming a power story, and possibly a legal one, after OpenAI reportedly concluded that its ChatGPT integration inside Apple Intelligence has not delivered the upside it expected.
As Bloomberg's Mark Gurman reported on May 14, Apple's two-year-old partnership with OpenAI has become strained, with the ChatGPT maker preparing possible legal action over what it sees as Apple's failure to make the integration meaningful enough for the business. That is a sharp turn from June 10, 2024, when Apple and OpenAI announced that GPT-4o-powered ChatGPT would come to iOS, iPadOS and macOS through Siri and Writing Tools.
At the time, the logic was obvious. Apple needed a credible generative AI partner because Siri had fallen behind ChatGPT, Gemini and Claude. OpenAI needed distribution, and there is no consumer technology channel quite like the iPhone. The deal came with careful privacy language: users had to approve requests before they were sent to ChatGPT, IP addresses would be obscured, and OpenAI would not store requests for users who were not signed in.
That structure also contained the problem. ChatGPT was not placed directly in front of every Apple user. It sat behind Siri, permission prompts, Apple's own interface choices, and a larger effort to keep Apple Intelligence branded as an Apple experience rather than an OpenAI one. Distribution is valuable, but only when the platform lets users feel it.
Apple's original pitch made ChatGPT sound like a powerful extension, not the center of Apple Intelligence. Siri could ask for permission to send a difficult query to ChatGPT. Writing Tools could use it for certain requests. Users could connect a paid ChatGPT account if they wanted extra features, but the default experience still belonged to Apple.
That may have protected Apple's brand and privacy positioning, but it limited OpenAI's ability to turn the relationship into direct customer growth. If a user receives an answer inside Apple's operating system, the credit often goes to the device, not the model provider behind one part of the response. That matters when OpenAI is trying to build consumer loyalty, justify enormous infrastructure spending and prove that ChatGPT is more than an app people visit when they remember to.
The commercial terms made the tension sharper. Earlier reporting said Apple was not paying OpenAI in cash for the integration, with distribution treated as the value. That can work if the partner receives major usage, account signups and status. It looks much weaker if Apple keeps the feature tucked inside a cautious, opt-in flow while it works on other model relationships.
Apple is moving toward more models
The relationship has also changed because Apple's AI strategy has changed. Apple has been exploring a more open Siri model, with reports that future versions could route requests to services such as Google Gemini, Anthropic's Claude and ChatGPT. That would make OpenAI less like Apple's chosen AI partner and more like one option in a menu controlled by Apple.
For Apple, that makes sense. The company does not want to be dependent on one outside model maker, especially in a market moving this quickly. It can use third-party models where they are strongest while continuing to present Apple Intelligence as the user-facing system. It also gives Apple leverage. If one partner is unhappy, another may be waiting.
For OpenAI, the same move weakens the original bargain. If the company accepted limited economics because the iPhone would make ChatGPT the default outside brain for Siri, a multi-model future changes the value of that position. It also makes any under-promotion today feel more consequential.
The companies are no longer only partners either. OpenAI's hardware push has made the relationship more complicated. In 2025, OpenAI moved to acquire Jony Ive's AI device startup in a deal valued near $6.5 billion, putting it closer to Apple's home turf. Bloomberg's latest report also points to Apple's frustration over OpenAI recruiting Apple hardware engineers for its device division. Apple protects its product talent fiercely, and OpenAI is now trying to build the kind of integrated AI hardware experience Apple would naturally see as strategic.
There is already legal smoke around the Apple-OpenAI tie-up from another direction. Elon Musk's xAI sued Apple and OpenAI in 2025, alleging that the ChatGPT integration and App Store treatment harmed AI competition. Apple has said the OpenAI arrangement is not exclusive, and the facts of that case are separate from OpenAI's own frustration. Still, it shows how quickly AI distribution on mobile devices has become a courtroom issue.
The bigger point is that AI deals are starting to look less like simple product integrations and more like battles over who owns the customer. OpenAI wants direct relationships, usage and data advantages. Apple wants control, privacy positioning and optionality. Those goals can overlap for a while. They rarely stay perfectly aligned.
For founders, the takeaway is blunt. Platform distribution can be powerful, but it is not the same as control. If Apple and OpenAI can turn a headline partnership into a dispute over effort, placement and value, smaller companies should assume every platform deal needs a plan for what happens when the platform's priorities change.
Also read: Meta Is Showing How AI Spending Can Break a Happy Workforce • A real Monet labeled as AI exposed the new trust problem online • Washington is turning AI guardrails into industrial policy
