Microsoft Copilot Joins CarPlay AI Chatbots, Gemini Stays Out

The race to put AI assistants in cars has quietly reached a milestone: Apple CarPlay now works with six major third-party AI chatbot apps, and Google Gemini is the only big-name holdout left standing. Microsoft Copilot is the newest arrival, joining OpenAI's ChatGPT, Perplexity, Grok, Meta AI, and Anthropic's Claude as in-car options for iPhone users931. For anyone tracking where AI copilots are actually being deployed rather than just demoed, the car dashboard has become one of the most contested new surfaces in tech13.

How Apple opened the dashboard

For years, Siri was the only voice assistant allowed inside CarPlay. Apple tightly restricted which app categories could appear on the in-car screen, citing driver distraction, which effectively locked out OpenAI, Anthropic, and Google from the vehicle entirely512. That changed when Bloomberg reported in early February 2026 that Apple planned to admit third-party chatbots, a shift one outlet described as the end of Siri's in-car monopoly1117.

The mechanism arrived with iOS 26.4. Apple's CarPlay Developer Guide added "voice-based conversational apps" as a supported app category, complete with a new voice control screen template that gives apps a standardized way to show visual feedback while a voice session is active1512. Developers have to request a specific entitlement and build against Apple's template — meaning every chatbot that shows up on the dashboard is, structurally, Apple's design with a vendor's brain inside it12.

The rules are strict in ways that matter for anyone thinking about serious copilot use. There is no wake word, so a driver cannot simply say "Hey ChatGPT" — the app has to be opened manually from the CarPlay screen. The assistants cannot control the car or the phone: no adjusting climate, no skipping songs, no reading messages, no access to live location or Maps2417. One reviewer's framing stuck: it is less an omniscient car intelligence than a well-informed passenger2.

Six deployments, six different bets

What's striking is how differently the six vendors have played the same opening. OpenAI moved first and moved big: ChatGPT arrived on CarPlay in April 2026, live globally, available on all plans at no extra cost, with an optional auto-start setting that launches voice mode every time the phone connects to the car3410. For safety reasons the app shows only conversation titles on the dashboard, never the text or images within them3.

Perplexity and Grok followed, with Grok representing xAI's first deployment outside the Musk ecosystem — a notable reach for an assistant that retains deeper, wake-word-activated integration inside Tesla vehicles, putting it in direct competition with ChatGPT across hundreds of millions of iPhones13. Early comparisons found the apps largely interchangeable in layout, though ChatGPT surfaces Projects and Grok alone lets drivers switch voices from the CarPlay screen18.

Meta AI became the fourth arrival at the end of August, but with a catch: the app appears in CarPlay while remaining in limited testing, and some users launching it get a message that the experience is not yet supported for their account68. Claude landed next, becoming the fifth major chatbot on the platform in early September, with a deliberately minimal interface offering a new voice conversation or a return to recent chats1920.

Then came Copilot. Microsoft's chatbot app now works through CarPlay on iOS 26.4 or later, and the in-car interface is voice-driven, pulls in recent conversations from the iPhone, and appears to require a paid subscription to use in the car9153132. That last detail is the most commercially significant of the whole rollout: it makes Copilot the first major chatbot to gate in-car access behind payment, at a moment when general AI pricing comparisons put Copilot's standard tier at twenty dollars a month and ChatGPT's entry tier at eight14.

Why the enterprise angle matters here

Read through an enterprise lens, the pattern across these six deployments tells a coherent story about how copilot distribution is evolving. The first wave of AI adoption was a browser tab on a work laptop. The second wave was embedded inside productivity suites. This is the third wave: ambient, voice-first surfaces that capture users during the dead time of a commute — precisely the hours knowledge workers spend between the home assistant and the office assistant.

Voice is the underrated variable. Every CarPlay deployment is voice-only by Apple's design, with minimal on-screen affordances and no text displayed while driving23. That constraint forces vendors to make their models genuinely good at spoken, hands-free, eyes-free interaction — a much harder test than a chat window, and one that maps directly onto how enterprises expect copilots to behave in field service, logistics, healthcare rounds, and manufacturing settings where hands and eyes are busy. In that sense, the car is functioning as a live proving ground for the interaction model enterprise copilot roadmaps are heading toward anyway.

The subscription split is the other signal. ChatGPT giving CarPlay away on every plan looks like a land-grab for habit — getting drivers to default to the assistant during unclaimed hours4. Microsoft gating CarPlay behind a subscription reads as the opposite bet: that in-car access is a premium feature worth paying for, consistent with a company whose AI business is overwhelmingly built on paid seats rather than free engagement. That is a defensible position when your buyers are enterprises rather than consumers, but it puts Copilot at a distribution disadvantage on the one screen where rivals are currently free.

There is also the gatekeeper question. Apple is not a neutral pipe here — it designed the template, controls the entitlement, and keeps Siri in the driver's seat, with third-party apps reachable only by tap and never by wake word2517. Every vendor building for CarPlay is accepting Apple's rules for how a conversational assistant behaves in a vehicle, which is a quiet form of standard-setting that enterprise buyers should watch: if the in-car pattern holds, Apple effectively defined the safety baseline for voice copilots in moving vehicles.

The Gemini question

The most telling wrinkle is Google's absence, and the reporting on it is worth reading closely because it diverges from earlier expectations. When Apple's plans first surfaced in February, coverage consistently named Gemini — alongside ChatGPT and Claude — as a service that would be accessible through CarPlay1511. Seven months later, Gemini is the only major consumer chatbot in this group without a dedicated CarPlay app1532.

None of the reporting explains why, and I'd commit to a reading rather than shrug: this looks less like a technical lag than a strategic choice. Google is the one vendor in this group that owns a competing in-car platform and a rival phone ecosystem, and it gains the least from strengthening an Apple-controlled surface it does not govern. A seven-month silence from the company with arguably the most resources to ship a template app is difficult to explain any other way. The counterpoint is that Claude, which was also named in the original February reporting, took until September to arrive, so late shipping is not unique to Google19. Still, the other five all showed up; only Gemini has not9.

Siri arrives on Monday

The competitive context is about to shift again. With iOS 27 leaving beta, CarPlay gains a dedicated Siri AI app, with Siri AI itself remaining in beta and rolling out in waves97. That means Apple's own assistant — newly chatbot-like, with conversation history — joins the same dashboard its rules govern, while every rival remains a tap-to-open guest rather than a wake-word resident617.

The likely endgame is not one winner but a default-plus-alternatives structure: Siri keeps the ambient role, and the six third-party assistants compete on model quality, voice performance, and price for the moments Siri is not good enough — which, if early user reports are any guide, is exactly how drivers are already treating it10. For the copilot market specifically, the takeaway is that deployment surface has become the differentiator. Six companies looked at the same rules and made six different calls on speed, price, and platform priority. Google's call is the only one still pending932.