TLDRocket
Sign in

ChatGPT can now send texts for you with new Apple Messages plug-in

TechCrunch Lucas Ropek

ChatGPT now plugs into Apple Messages, so it can sort, draft, send, and even delete texts. OpenAI says it runs locally, but letting an AI speak for you is still a privacy tripwire.

Based on reporting by TechCrunch, Lucas Ropek — read the original for the full story.

Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error

OpenAI has added an Apple Messages plug-in to ChatGPT, letting users hook their Messages inbox directly into the chatbot. The pitch is simple: let the model sort through conversations, analyze them, edit them, and act on them without making you bounce between apps.

The plug-in is not just for casual texting. OpenAI says it also works with Codex and ChatGPT Work, which means the same Messages connection can be used in a professional setting too. In a short ad for the feature, a user asks ChatGPT to suggest follow-up texts based on messages received the day before.

And the tool can go further than suggestions. Users can ask ChatGPT to delete messages, draft replies, send messages on their behalf, or dig up something buried in message history. That last part is the kind of convenience that sounds harmless until you remember how much of a person's life lives inside a chat thread.

Privacy is the obvious issue here. OpenAI told Bloomberg the plug-in runs locally on a user's machine and does not build an index of all someone’s messages, though the company did not make the mechanics of that claim fully clear. TechCrunch has asked OpenAI for more information.

OpenAI is also warning people to pay attention when message sending is turned on. The company says users should avoid persistent approval, because it removes the final chance to review a message before ChatGPT sends it as the user. That is a fairly elegant way of saying: if the bot is getting the keys, don’t walk away from the car.

My take — AI-written commentary, not fact-checked reporting

This is the same old AI play: sell convenience first, explain the trust model later. A Messages plug-in sounds useful right up until the part where your chatbot is one bad prompt away from speaking for you. The industry keeps calling this productivity; normal people would call it handing over the phone and hoping for the best.

Read more about this at: TechCrunch

Related stories

The daily briefing

Every AI story that matters, in your inbox by 8am.

TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the day in two minutes. Follow companies and topics for alerts, or get the briefing in Slack. Free, no spam, unsubscribe anytime.