ChatGPT can now send your texts through Apple Messages
OpenAI just wired ChatGPT into Apple Messages, and the move says as much about where assistants are heading as it does about iMessage.
OpenAI launched a ChatGPT plugin that can read, draft, and send your Apple Messages. The new integration connects a user's Messages inbox to the chatbot, letting it sort, summarize, and rewrite conversations, suggest follow-ups based on the previous day's threads, and — with the user's go-ahead — compose and send replies on their behalf. It also works inside Codex and ChatGPT Work, pushing the feature from personal convenience into the workplace. OpenAI is explicit that you should keep a human in the loop: the company warns against flipping on persistent approval, because that "removes your final chance to review a message before ChatGPT sends it as you." The privacy trade-off is real — you're handing an AI lab a window into your most personal conversations — though OpenAI told Bloomberg the plugin runs locally on the user's device and does not build a full index of their messages.
Twin1 AI raised $20 million to give every knowledge worker a persistent AI "twin." The San Mateo startup, founded by ex-Eigen Technologies leaders including CEO Lewis Liu, builds a private model of an individual's expertise, judgment, and work context that can answer questions and act on that person's behalf across Slack, Teams, Outlook, Gmail, and SharePoint. The pitch is governance, not just automation: six layers of rules- and AI-based controls gate what a twin will surface, and a "Twin Network" lets coworkers route requests to the right person's twin with permission awareness. Twin1 says named customers like Linklaters and Orrick already report the platform handling 30–50% of the communications work their knowledge workers would otherwise do. Bessemer, Tribeca, and Aramco Ventures co-led the round, and the company plans a self-service tier after its enterprise product.
What to watch: whether Apple pushes back on an OpenAI layer sitting inside iMessage, and whether "digital twins" of employees survive contact with real corporate IT and compliance teams.
Should an AI be allowed to send messages in your name, or is that a line assistants shouldn't cross? Tell us in the comments.
