ChatGPT Can Now Read and Send Messages on Mac — But the Privacy Cost Is Raising Eyebrows
ChatGPT is becoming far more deeply integrated with the Mac, and its newest capability could change how people handle their everyday conversations. OpenAI has introduced an Apple Messages integration that allows ChatGPT to search conversations, analyze messages, draft replies and even send messages through the Messages app.
The feature is designed to turn ChatGPT into more than a chatbot sitting alongside your apps. With the appropriate permissions, it can work directly with conversations stored in Apple’s Messages app, giving users a way to find information buried inside old chats or automate routine replies. The feature currently works through ChatGPT Work and Codex on desktop rather than ordinary ChatGPT conversations.
ChatGPT Can Actually Work With Your Messages
The new integration supports iMessage, SMS and RCS conversations stored in Apple’s Messages app on Mac. Users can ask ChatGPT to search through conversations, summarize discussions, identify information and prepare responses.
It can also go one step further: ChatGPT can send a message through Apple’s Messages app on the user’s behalf after the necessary permissions have been granted. OpenAI has demonstrated use cases such as finding conversations that need a follow-up and helping users respond based on other information available on their computer.
That makes the feature potentially useful for people who treat Messages as an informal task manager. Instead of scrolling through months of conversations looking for a forgotten detail, users can simply ask ChatGPT to find it.
The Feature Requires Significant Mac Permissions
This is where the story becomes considerably more complicated.
To interact with Messages, ChatGPT requires several macOS permissions, including Full Disk Access, access to contact information and automation-related permissions. Full Disk Access is particularly notable because it is a broad system permission rather than an authorization limited specifically to the Messages database.
That requirement has immediately raised questions about how much access users are effectively giving an AI assistant in exchange for messaging convenience.
The important distinction is that having the permission does not mean ChatGPT is constantly reading every message automatically. OpenAI says the Messages integration is activated by the user and operates locally on the Mac. The company has also said that it does not create an index of a user’s messages.
Still, users should understand exactly what permissions they are granting before enabling the integration.
There Is Still a Human Approval Step
Sending messages through an AI assistant introduces another potential problem: an AI-generated response could be sent to the wrong person or contain something the user never intended to communicate.
OpenAI therefore includes an approval step before messages are sent. However, users can enable persistent approval, which removes that final opportunity to review individual messages. OpenAI itself recommends being cautious with this setting.
For something as personal as texting, that distinction matters.
A typo in a generated email might be embarrassing. An incorrectly generated personal message sent to a friend, family member, colleague or partner could have much bigger consequences.
The Privacy Debate Has Already Started
The arrival of the Messages integration has sparked a broader debate about where the line should be drawn between AI assistance and personal privacy.
On one side, the feature could be extremely convenient. Imagine asking ChatGPT to find the restaurant someone recommended several months ago, identify conversations you forgot to answer or summarize an important discussion without manually searching through hundreds of messages.
On the other side, Messages can contain some of the most personal information stored on a computer. Giving an AI assistant the ability to access that environment inevitably creates concerns about permissions, accidental disclosure and what could happen if the integration were misused.
Security researchers and commentators have also pointed to potential risks involving malicious instructions embedded within content that an AI system encounters while operating across applications.
This Is Bigger Than Just iMessage
The significance of the update goes beyond simply allowing ChatGPT to send texts.
OpenAI has been steadily pushing ChatGPT toward becoming an agent that can understand and operate across a user’s digital environment. Its recently introduced Computer History feature, for example, allows the Mac application to maintain a timeline of certain user activity so ChatGPT can help users resume workflows and retrieve previously accessed information.
The Messages integration fits into that larger direction.
Instead of asking an AI a question and receiving an answer, users are increasingly able to tell it what they want accomplished and have it interact with the applications required to complete the task.
Convenience vs. Control
The real question surrounding this update isn’t whether ChatGPT can read and send messages. It can.
The bigger question is whether users are comfortable giving an AI system enough access to make that capability useful.
For people who carefully control application permissions, the requirement for Full Disk Access may be a major reason to think twice. For others, the ability to search years of conversations and automate repetitive communication could make the trade-off worthwhile.
Either way, the update represents another significant step toward AI assistants becoming active participants in our digital lives rather than simply tools we open when we need an answer.
And as ChatGPT gains access to more of the applications we use every day, privacy and user control are likely to become just as important to the AI experience as intelligence itself.
What the New ChatGPT Messages Integration Can Do
– Search conversations in Apple Messages
– Work with iMessage, SMS and RCS chats
– Analyze and summarize conversations
– Draft replies
– Send messages through the Messages app
– Help users find information buried in old conversations
– Work with other information available through supported Mac workflows
Geeksterr Take: ChatGPT controlling Messages is undeniably impressive, but it also represents the kind of AI advancement where convenience needs to be balanced carefully against privacy. The technology is moving from answering questions to taking actions — and that makes the permissions we grant it more important than ever.
