ChatGPT Can Now Read and Send Apple Messages on Mac

Key Highlights of NEWS

• OpenAI has added an Apple Messages plugin to the ChatGPT desktop app for macOS.
• The plugin can read, search, summarise, analyse, draft and send conversations through Apple's Messages app.
• It works with iMessage, SMS and RCS conversations stored on an Apple Silicon Mac.
• ChatGPT asks for user approval before sending a message by default, while OpenAI warns against enabling persistent approval for sensitive conversations.

AI assistants are increasingly moving beyond text generation to perform tasks across everyday applications. Explore Artificial Intelligence courses to build practical skills in AI, Generative AI, and emerging technologies shaping the future of work.

What can ChatGPT do with Apple Messages?

OpenAI has introduced an Apple Messages plugin for the ChatGPT desktop app on macOS, allowing the AI assistant to work directly with conversations stored in Apple's Messages app. The update rolled out on August 20, 2026, and expands ChatGPT from generating text to taking actions inside a user's messaging workflow.

With the plugin enabled, users can ask ChatGPT to find specific conversations, search through message history, summarise chats and analyse previous exchanges. It can also draft replies and send messages through the Messages app. TechCrunch reports that the plugin can even delete messages when instructed.

The integration covers iMessage, SMS and RCS, giving ChatGPT access to the different types of conversations that can appear in Apple's Messages application. However, the feature does not let users chat with ChatGPT remotely through Messages, and it does not work inside regular ChatGPT chats.

As AI systems gain the ability to interact with applications and make decisions within business workflows, leaders need to understand their strategic and operational implications. Explore the Chief Technology Officer & AI Leadership Programme to develop the expertise needed to lead AI-driven transformation responsibly.

Which Macs and ChatGPT modes support the feature?

The Apple Messages plugin works on Apple Silicon Macs. Business Standard reports that the feature is available across ChatGPT plans in the desktop app, but users need a Mac powered by Apple Silicon to use it.

OpenAI has also made the integration available through ChatGPT Work and Codex on desktop, extending its potential use beyond personal conversations. For example, users could search older work-related messages, identify follow-ups or prepare responses based on earlier discussions.

The integration requires users to grant macOS permissions related to automation, contacts and Full Disk Access. These permissions allow ChatGPT to interact with Messages, identify contacts and search the stored message history.

That permission model is important because the feature does not simply generate a response in isolation. It gives an AI system access to the context contained inside a user's existing conversations.

Why does the Apple Messages integration raise privacy concerns?

The biggest concern is the amount of personal information an AI assistant can potentially access. Messages can contain private conversations, contact details, addresses, financial information and other sensitive information, making access to message history materially different from asking an AI to write a generic text.

TechCrunch reported that OpenAI told Bloomberg the plugin runs locally on the user's machine and does not create an index of all the user's messages. However, TechCrunch noted that the precise implications of this setup were not immediately clear.

Message sending also includes a safety mechanism. By default, ChatGPT asks users to approve both the message and its recipients before sending. Users can enable persistent approval for a particular chat, but OpenAI warns that doing so removes the final opportunity to review a message before ChatGPT sends it.

The update illustrates a broader shift in AI development: assistants are moving from answering questions to interacting with the applications and data people already use. That makes AI agents more useful, but it also makes permissions, privacy and user oversight increasingly important.