Here’s how ChatGPT can read and send your Apple messages
The opt-in plugin needs Full Disk Access and works only inside ChatGPT Work and Codex
OpenAI has launched a plugin that lets ChatGPT read, search, draft and send messages through Apple's own Messages app on Mac, and as Bloomberg notes, the move is already raising privacy and security questions.
The feature works with iMessage, SMS and RCS conversations, giving the chatbot a level of access to personal communication it hasn't had before.
How ChatGPT read Apple messages?
The plugin is supported in every ChatGPT desktop plan on macOS but is only functional in ChatGPT Work and Codex but not in regular ChatGPT conversations.
In an example from OpenAI itself, a user asked ChatGPT to locate the missing conversations from the previous day and then got an assistant to write and respond using the Messages app.
The plugin could be used to summarise conversations, look for any spam that needs to be deleted, or search for birthdays to put into the calendar. At present, the plugin is only supported in Apple Silicon Macs.
Users must approve ChatGPT's access to their Mac's on-device Messages history during setup and then separately grant Full Disc Access through macOS System Settings, along with permission to see contact names and use automation tools.
That multi-step chain means, as Bloomberg points out, nobody ends up with this enabled by accident. Sending a message still requires manual approval by default, and OpenAI explicitly warns against turning on persistent approval, since doing so removes the final check before a message goes out under a user's name.
In 2024, the company repeatedly disabled Beeper Mini's access to iMessage until the reverse-engineered chat app's developers stopped trying to restore it.
The timing here is notable too: Apple sued OpenAI in July, accusing the company of a scheme to extract confidential product information by hiring away Apple employees, and Apple reportedly dropped OpenAI in favour of Google's Gemini for its own AI features earlier this year.
-
Meta, TikTok and X challenge UK watchdog over safety data: What to know
-
Norway proposes temporary ban on AI glasses in public spaces over privacy fears
-
OpenAI CEO Sam Altman argues global society must accept AI downsides to secure technological benefits
-
Google to downgrading Gemini for free, AI Plus users
-
WhatsApp login block: Update your OS or lose access
-
Qualcomm signs deal to license Huawei patents for 5G, AI
-
Trump's Super Intelligence Force: What to know
-
OpenAI Sam Altman warns of new safety risk: ‘AI becoming religious force’