ChatGPT adds Apple Messages plug-in

OpenAI’s new plug-in lets ChatGPT search, edit, and send texts from Apple Messages, while the company says message data stays on the device.

ChatGPT can now connect to Apple Messages through a new plug-in from OpenAI, giving users a way to work with text conversations inside the chatbot. The feature is aimed at both personal and professional use, and it raises immediate privacy questions because it can handle sensitive message data. OpenAI says the plug-in can sort, analyze, and edit messages, draft replies, and even send texts on a user’s behalf. As IT-PUB News notes, the company is also tying it to Codex and ChatGPT Work, suggesting this is meant to be more than a simple personal messaging add-on.

ChatGPT can now work with Apple Messages

The new plug-in lets users connect their Apple Messages inbox to ChatGPT. OpenAI says the goal is to make message threads easier to manage, especially when they become hard to track manually.

A short promotional video shows a user asking ChatGPT to suggest follow-up messages based on the previous day’s conversations. OpenAI also says people can ask the chatbot to search for information buried in their message history.

That makes the plug-in more ambitious than a basic writing aid. It is not limited to helping with a reply in the moment — it can also look back through past conversations and work with message content already stored on a device.

OpenAI says message handling stays local

Because the feature deals with private conversations, OpenAI was pressed on how it handles data. In response to TechCrunch, the company said ChatGPT does not build a full index of a user’s messages.

OpenAI also said the Messages plug-in runs locally on the user’s device. According to the company, ChatGPT can read messages only when the user makes a specific request.

To set up the plug-in, users must enable Full Disk Access. OpenAI says that gives the app the ability to read and write files on the device.

In a later email, the company added that ChatGPT desktop stores conversations by default on the user’s computer. OpenAI also said message content is stored locally and is not saved to its servers.

If a user chooses to save a conversation in the cloud, OpenAI said the retention policy is the same as for other content stored there.

Sending texts adds another point of concern

The most sensitive part of the feature may be sending messages, not just reading them. OpenAI says users should watch what ChatGPT is doing and warns against turning on persistent approval.

The company says that doing so “removes your final chance to review a message before ChatGPT sends it as you.”

That warning gets to the trade-off at the center of the plug-in. It may save time and cut down the effort of replying, but it also creates a risk that users could give up direct control over what gets sent if they lean too heavily on automation.

For everyday users, that could mean faster replies and easier sorting of cluttered inboxes. For professionals, OpenAI’s mention of ChatGPT Work and Codex suggests the tool could become part of a broader communication workflow. In both cases, though, the feature depends on access to highly personal data.

Full Disk Access leaves privacy concerns in place

OpenAI has described some protections, but the overall privacy setup is still not fully clear. That is likely to draw attention from users being asked to grant broad permissions in exchange for convenience.

The company’s explanation that data stays local on the user’s computer is meant to reassure people worried about cloud storage. Even so, the need for Full Disk Access — and the ability to read and write files on the device — may make some users cautious.

The plug-in shows ChatGPT moving beyond answering prompts and into taking direct action inside everyday apps. Here, that means access to one of the most personal digital spaces many people use: their messages.