Apple announces Apple Intelligence with OpenAI inside

At its Worldwide Developers Conference, Apple announced Apple Intelligence, a suite of generative-AI features built into iOS 18, iPadOS 18 and macOS Sequoia, and disclosed that OpenAI would provide the system’s outside “knowledge” layer through a ChatGPT integration built on GPT-4o. Most Apple Intelligence functions — writing tools, notification summaries, an overhauled Siri, image generation — run on Apple’s own on-device and cloud models. When a request exceeded what Siri could handle itself, it would ask the user’s permission to forward the query to ChatGPT.

Apple set the terms of the integration deliberately narrowly. ChatGPT access would be free, require no OpenAI account, and, Apple said, would not be logged by OpenAI as user data. The company also said it intended to add other AI providers over time, framing OpenAI as the first partner rather than an exclusive one — an arrangement read at the time as protecting Apple’s negotiating position and its privacy messaging simultaneously.

The deal mattered less for what it built than for what it distributed. For Apple, it filled a capability gap without the years and capital cost of training a frontier model in-house, while keeping the flagship Siri experience nominally Apple’s own. For OpenAI, it was the widest single distribution channel any AI company had secured, placing ChatGPT as a fallback option across a large share of the world’s smartphones without OpenAI having to build the integration into every app itself. The financial terms of the arrangement were not disclosed. The announcement also sharpened a rivalry: Elon Musk, who had no equivalent deal for xAI, said he would restrict Apple devices at his companies over the OpenAI integration, calling it an unacceptable security risk — a threat he did not carry out.