OpenAI lets users text dots, its ChatGPT agents, in preview in the US
IA en un minuto newsroom · Editor: Jon Elgezabal
In 30 seconds
OpenAI is opening text messaging to dots, its ChatGPT agents, as a beta; only people on ChatGPT Pro in the United States get it for now. The agent answers on iPhone through iMessage and on Android through RCS, taking text, photos or voice notes, and it may start the chat to send alerts. OpenAI has not set dates for other plans or countries.
ChatGPT agents are starting to arrive in the messaging app on phones. OpenAI has begun rolling out text messaging for dots, its agents that work in the background. According to OpenAI's what's-new note, which covers what shipped between October 26 and 30, 2026, the service is arriving in beta for ChatGPT Pro subscribers in the United States and works through iMessage on iPhone or RCS on Android.
Users can hand a task to their dot or ask it to keep watch over something, using text, photos or voice messages, and the agent can also message them by itself when it has news. According to the channels documentation, it is switched on from the dot's profile, under the Texting option: the phone number is verified and the contact card the agent sends is saved. On Android, RCS must be turned on. The dot draws on the context of ChatGPT conversations and of the connected channels and plugins to decide where to reach the user, and it supports only one-to-one conversations.
OpenAI says only the user can message their dot and the dot can only message the user. For now there is no timetable for other plans or other countries. According to Diario TI, citing Digital Trends, OpenAI introduced dots on September 29, and until now they could take instructions in ChatGPT, Slack and Microsoft Teams, and through a widget and shortcuts on iOS.
Why it matters · analysis and opinion
The change is not in what the agent does but in where it is found: it moves from a window you have to open to the messaging app people already check every day, and from there the agent can be the one to start the conversation. That removes friction when asking it for things, but it also raises the risk of too many alerts, which is why OpenAI lets users set rules so it writes only when something needs attention. For now it is a trial rollout, limited to one plan and one country, and the company has not said when it will reach others, so a European business cannot count on it yet. The sensible step is to follow it closely and think about which monitoring or reminder tasks it would hand to an agent like this, and which approvals it would still want to give personally.
Official source: OpenAI · Written with the help of AI: how we make the news
