OpenAI is rolling out a Messages integration for its ChatGPT macOS app that lets the assistant read, search, summarize, draft and send iMessage conversations after users grant the necessary permissions. The feature brings ChatGPT directly into one of the Mac's most personal communication environments, joining the assistant's existing hooks into Notes, Xcode and other native apps.
What The Integration Does
After a user grants access, ChatGPT can pull threads out of Messages, generate summaries or reply drafts, and send outgoing texts on the user's behalf. OpenAI says the Messages integration processes conversation data locally and does not create a separate index of the user's Messages history. That guardrail is important: unfiltered access to private chats would open a large surface for prompt injection or accidental data exfiltration if third parties embed instructions inside messages.
ChatGPT Keeps Moving Into The OS
The Messages integration is another data point in a broader shift from standalone chatbots toward assistants that act across everyday applications. Cloudflare's Kitesurf agent browser and Alibaba's Qwen UI agent targeted similar automation surfaces, and Apple's own Visual Intelligence and Siri upgrades are set to arrive this fall. The trajectory is clear: AI is becoming an operating layer over existing apps rather than a destination product.
Enterprise And Security Implications
Because Messages sync across iCloud, an integration on the Mac effectively reaches phone-based conversations too. Security researchers are likely to probe how OpenAI enforces its local-processing claim and how granular the permission model is per contact and per thread. Enterprises with managed macOS fleets will want MDM controls before enabling the feature. For now, users see it as an optional opt-in inside the ChatGPT app.
Reporting based on coverage from 9to5Mac, TechStartups and OpenAI communications.