I Gave ChatGPT Access to My iMessage Conversations
Tech journalist tests the new Apple Messages plugin on the ChatGPT macOS desktop app, highlighting chat history retrieval and privacy features.
Stock photo for illustration only, not from the actual event
- The ChatGPT desktop app on macOS now features an Apple Messages plugin for iMessage integration.
- All processing occurs locally on your Mac without sending chat data to cloud servers.
- Users can prompt the AI naturally to send messages, search archives, or analyze chat history.
David Nield, a technology journalist with over 20 years of experience covering gadgets and apps, has tested the new Apple Messages plugin on the ChatGPT macOS desktop app. This new capability extends the desktop client's existing local file, email, and screenshot search features directly to Apple's messaging platform.
While granting an AI access to personal chat logs prompts typical privacy concerns from users on platforms like Reddit—with reactions labeling it spy software—many people already rely on artificial intelligence to compose and interpret messages. Competitor Claude also offers a similar Apple Messages plugin for users.
OpenAI confirmed to TechCrunch that the plugin performs all processing directly on the user's Mac, avoiding any transmission of chat data back to cloud servers. Consequently, these iMessage conversations will not show up in ChatGPT web or mobile applications, and the desktop app does not build a complete index of all text messages.
Mac users can set this up by downloading the desktop app, navigating to the Plugins tab, searching for "messages," and clicking Install next to Apple Messages. Prompts can then be executed using natural language, the @Messages mention, or through the plugin list in the prompt box.
Stock photo for illustration only, not from the actual event
During practical testing, Nield found that ChatGPT effectively drafts natural messages, such as politely declining a dinner party invitation. Users retain full control with the option to review and approve messages before they are sent, or configure them for automatic sending in selected conversations.
Processing chat data entirely on the local device represents a significant shift in addressing AI privacy challenges. Because text messages contain highly sensitive personal information, keeping computations local minimizes cloud server vulnerability risks, though users must still carefully weigh convenience against personal privacy boundaries.
The most practical utility lies in chat history retrieval and analysis. Nield used the AI to locate event arrangements, delivery details, and even carrier text messages to track phone bill increases over time, with the system returning exact monetary increases, percentage changes, and specific dates.
Despite its usefulness, potential risks remain regarding over-reliance on AI outputs. The technology can be utilized to assess someone's mood over time, draft relationship-rekindling messages, or fabricate excuses to get out of work, pushing the tool from convenient assistance into uncomfortable territory.
Source: Lifehacker
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment