The release enables ChatGPT to read, write, and send iMessages on Mac, allowing the AI to search conversations and summarize content within Messages, a capability that raises privacy concerns for Apple given its privacy stance. OpenAI says the feature requires user consent, runs on-device, and leverages macOS tools, with setup involving multiple opt-ins and permissions such as Full Disk Access. The rollout occurs amid a high-profile Apple lawsuit accusing OpenAI of theft of trade secrets, underscoring a fraught relationship between the companies. The development could pressure Apple’s privacy branding and positions AI-assisted messaging as a competing force to forthcoming Siri AI, signaling a broader shift in on-device AI integration.
Dive Deeper:
The feature operates with ChatGPT on the Mac and enables actions inside Apple’s Messages app, including reading, writing, sending texts, searching conversations, and generating summaries of message content.
OpenAI maintains that the tool requires explicit user consent and does not construct an index of all messages; the plug-in runs locally on the Mac and relies on built-in OS tools like AppleScript and Accessibility.
Setup involves several opt-in steps in both the ChatGPT app and macOS, including a permissions screen that indicates access to on-device message history and additional privacy settings adjustments.
Users must grant Full Disk Access and permit ChatGPT to access contact names and automation tools, a process comparable to enabling other AI automation software.
Apple’s privacy-focused brand context is highlighted by concerns that a bridge to OpenAI could expose user data to external servers, potentially clashing with its messaging privacy promises.
The move is notable given a legal action where Apple accuses OpenAI of widespread theft of trade secrets and confidential information across multiple levels of OpenAI’s organization, filed in federal court in Northern California.