OpenAI activated a controversial ChatGPT desktop feature on Thursday that gives the AI assistant direct control over Apple Messages on the Mac. Once users opt in through both the ChatGPT app and macOS system settings — including granting Full Disk Access, approving contacts, and enabling automation permissions — the assistant can read, search, draft, and send iMessage, SMS, and RCS texts on the user’s behalf. The integration also works inside ChatGPT Work and Codex, extending the same message-handling capabilities to professional accounts.
The plug-in uses existing macOS tools such as AppleScript and Accessibility rather than exploits. OpenAI says it runs locally on the Mac and does not create a full index of message history. By default, ChatGPT asks for approval before sending each message, but the company warns that enabling the optional “persistent approval” mode removes the user’s final chance to review a message before it is sent.
The launch arrives during an active legal clash: Apple sued OpenAI weeks ago, alleging theft of trade secrets and confidential information about upcoming products. Earlier in 2026, Apple dropped OpenAI in favor of Google’s Gemini for its built-in assistant, and OpenAI had reportedly hired outside attorneys to examine a lawsuit against Apple over the Siri-ChatGPT deal. Apple has not commented on the new Messages plug-in, leaving open whether it will block the integration as it previously did with Beeper Mini.
Privacy advocates have raised concerns about third-party AI access to sensitive communication data. OpenAI points to the multiple opt-in steps and local processing as safeguards, but it has remained vague about exactly what data is processed or retained. The feature follows OpenAI’s expansion into Apple Health data on iPhone and iPad, and it arrives ahead of Apple’s expected Siri refresh that would also search on-device content including messages.
