This article was AI-generated as part of an experimental historical-content project. The date reflects the period being analyzed rather than the date the article was originally written.
Most of the commentary on Apple’s WWDC keynote has been about whether Apple is late to AI. For anyone who thinks about how companies and people get described, a different detail matters more. A growing share of questions about you may soon be asked out loud, to a phone, and answered without a results page anywhere in sight.
What Apple announced
On Monday Apple introduced Apple Intelligence, a set of generative features built into iOS 18, iPadOS 18 and macOS Sequoia. Much of it runs on the device, with harder requests sent to what Apple calls Private Cloud Compute. It arrives in beta this fall, in U.S. English, on the iPhone 15 Pro and on Macs and iPads with M1 chips or later.
The part that caught my attention is the ChatGPT integration. According to Apple and OpenAI, Siri will be able to tap ChatGPT “when helpful.” Users are asked before a question is sent, and Siri then presents the answer directly. It is powered by GPT-4o, free without an account, and due “later this year.” OpenAI says those requests will not be stored and IP addresses will be obscured. None of it is live yet.
Follow the chain
It helps to trace the path an answer would take. Someone on an iPhone asks Siri about a company they are about to do business with. Siri decides the question needs broad world knowledge. A prompt asks permission to use ChatGPT. The user taps yes. ChatGPT composes an answer from what it learned from the web, which means from news coverage, Wikipedia, company sites and everything else it was trained on. Siri reads it back.
Three brands are involved in that one answer. Apple owns the interface and the voice. OpenAI owns the model. The sources that shaped the content, possibly including your own press materials and a five-year-old article about a lawsuit, are the least visible of all.
That last point is the reputation issue. On Google, even an AI Overview sits on top of a page of links that a curious person can scroll through. A spoken or on-screen Siri answer has no scroll. The answer is the whole experience.
The permission prompt is also a routing decision
Apple presents the prompt as a privacy feature, and it is one. It is also a routing decision. Some questions will stay with Apple’s own models and some will go to ChatGPT. Apple has not said exactly which kinds of questions qualify, so it would be speculation to guess where a question about a specific company lands. What is clear is that the answering layer is becoming modular, and the person asking may not be thinking about which engine replied.
There is a quieter piece of the announcement too. Apple Intelligence will summarize emails in the inbox instead of showing their first lines, and summarize notifications on the lock screen. A journalist reading your pitch, or an investor reading your update, may first see the system’s summary of it.
What follows for communications teams
None of this is in anyone’s hands yet, so the right response is preparation rather than alarm.
Ask ChatGPT the questions a stranger might ask Siri about your company and your executives. That is the closest preview available of what the integration would draw on, and earlier posts here have argued that what a model says when nobody is watching matters more than the demo.
Write the first lines of announcements and emails so they survive being summarized. If the key fact is in the fourth paragraph, assume a summary might not reach it.
And widen the monitoring map. Google, Wikipedia, Gemini, Copilot and ChatGPT already compete to summarize you first. The phone’s assistant is about to join them, with the reach of the iPhone behind it once the supported devices spread.
The lock screen used to show notifications. It may soon show conclusions.