ChatGPT Can Now Read and Send Your iMessages — And People Are Not Happy About It

OpenAI has quietly rolled out one of its most privacy-sensitive integrations yet: the ability for ChatGPT on Mac to read, write, search and send messages through Apple's iMessage service. The move lands at a moment of active legal conflict between the two companies and directly encroaches on territory Apple has been building Siri to own — creating a collision between two of the most powerful AI platforms competing for control of the iPhone user experience.

Key Points

• ChatGPT on Mac can now read, write, search and send messages through Apple's iMessage app, as well as summarise conversations
• OpenAI says the feature runs locally on Mac using AppleScript and Accessibility tools, requires explicit user consent and does not create a full index of messages
• Users must grant Full Disk Access in macOS System Settings and provide permissions for contact names and automation tools
• Apple sued OpenAI last month, accusing the company of stealing information about upcoming products
• The integration threatens to compete directly with Siri's planned ability to search and act on message content — a feature Apple considers a key differentiator
• Apple did not respond to requests for comment

What the Integration Does

The feature, available through ChatGPT's Mac application, gives OpenAI's AI system the ability to interact with Apple Messages in ways that go well beyond passive reading. ChatGPT can read existing messages, compose new ones, send them, search through message history and generate summaries and contextual information about conversations.

The technical implementation uses existing macOS tools — specifically AppleScript and the Accessibility framework — to interact with the Messages application. OpenAI emphasised that the plug-in runs locally on the Mac rather than uploading a user's message history to OpenAI's servers, and that it does not create a persistent index of all messages on the device.

That local processing framing is a key part of OpenAI's positioning around the feature's privacy implications. The distinction between an AI system that processes messages on-device versus one that transmits them to a cloud server is meaningful — but it does not fully address the concern that a third-party AI system now has programmatic access to one of the most sensitive data stores on a user's device.

The Consent Architecture

OpenAI has built multiple consent checkpoints into the integration in an apparent effort to address the privacy concerns it anticipated.

When a user first sets up the integration, a permissions screen appears within the ChatGPT app explaining that the computer's on-device message history will be accessed. Users must also make changes within macOS System Settings — including enabling Full Disk Access, a permission that grants an application access to protected areas of the file system that most applications cannot reach without explicit authorisation.

Additional permissions are required for access to contact names and automation tools. The process mirrors the setup flow for other AI-based automation software including Codex Computer Use, suggesting OpenAI has built the Messages integration on the same underlying access framework it uses for broader computer automation features.

The multi-step consent architecture is a defensive measure — the more explicit and numerous the consent steps, the harder it is for critics to argue that users were not adequately informed. But it also creates a practical reality: most users who complete a lengthy permission sequence to enable a feature do not subsequently audit what that feature is doing with the access it has been granted.

Why Apple Has Every Reason to Be Concerned

Apple's position on user privacy is one of its most commercially significant brand commitments. The company has built substantial marketing capital around the premise that its devices and services protect user data in ways that competitors do not — and iMessage, with its end-to-end encryption and strong privacy protections, is one of the clearest expressions of that commitment.

A third-party AI system that can programmatically read, write and send iMessages — even with user consent and local processing — represents a departure from the controlled ecosystem Apple has maintained around its messaging infrastructure. The company shut down Beeper Mini, an application that extended iMessage functionality to Android devices, not because it was technically breaking anything but because it involved third-party interaction with Apple's messaging infrastructure that Apple had not authorised.

The OpenAI integration operates through sanctioned macOS APIs — AppleScript and Accessibility — rather than through an unauthorised protocol implementation like Beeper Mini. But the principle that Apple controls how its messaging platform is extended and integrated remains, and an unsanctioned third-party AI with read-write access to iMessage is a significant challenge to that principle.

Apple's silence in response to press inquiries is notable given how quickly the company typically moves to address perceived threats to its messaging infrastructure.

The Siri Competition

The integration's competitive implications extend beyond the immediate privacy debate. Apple has been positioning an enhanced Siri — capable of searching through messages, emails and other device content to provide contextual AI assistance — as a key differentiator for its own platform AI offering. The ability to comb through a user's iMessage history and provide summaries, context and action on that content is precisely the capability Apple intends Siri to own as a native, deeply integrated feature.

ChatGPT now does that too — and it does it before Apple's enhanced Siri has been fully rolled out to users. That sequencing matters commercially. Users who establish ChatGPT as their primary AI interface for message management before Siri offers comparable functionality may be significantly less motivated to adopt Apple's version when it arrives.

OpenAI has already extended ChatGPT's reach into Apple's ecosystem through integration with the Health app on iPhone and iPad, where iOS and iPadOS require users to specifically approve data sharing with ChatGPT. The Messages integration on Mac adds a more sensitive and strategically significant data category to ChatGPT's access scope.

The Legal Context

The rollout arrives during active litigation between the two companies. Apple sued OpenAI last month, accusing the AI provider of stealing information about upcoming products. The nature of the alleged theft and the specific products involved have not been fully detailed in public filings, but the lawsuit signals a relationship between the two companies that has moved from partnership — Apple integrated ChatGPT into Siri through a deal announced in 2024 — into active legal conflict.

Rolling out an integration that gives ChatGPT access to iMessage while simultaneously facing a lawsuit from Apple is either a demonstration of how independent the two companies' product development processes are from their legal situations, or a deliberate signal about OpenAI's willingness to operate in Apple's ecosystem without Apple's blessing.

ChatGPT's existing Mac integrations — with Xcode, Notes and Terminal — have not produced comparable friction with Apple. The Messages integration is different in kind: it touches encrypted communication content, competes directly with Siri's planned capabilities and sits at the centre of Apple's privacy brand identity in a way that development tools and note-taking integrations do not.

Sources

CNBC reporting on ChatGPT iMessage integration rollout, August 2026. OpenAI response to privacy questions regarding Messages integration, August 2026. Apple lawsuit against OpenAI reference, July 2026. Beeper Mini Apple shutdown reference, cited for historical context of Apple's approach to third-party iMessage integration. OpenAI Health app integration reference, iOS/iPadOS consent requirement.