OpenAI adds iMessage and RCS texting to Dot agents in a preview for US Pro users

OpenAI has begun rolling out a preview that lets US Pro users text their Dot agents over iMessage and RCS, according to Engadget. The company says users can send texts, photos and voice memos to a Dot, and the Dot can text back proactively with updates on tasks it has been given.

Engadget reports that OpenAI promised texting when it debuted Dot agents late last month. The preview arrives nearly a month later.

What the feature does

OpenAI describes the feature in a post on the official ChatGPT account on X: users can give a Dot "a task to take off your plate or something to keep an eye on," using text, photos or voice notes. The Dot then texts the user proactively with relevant updates.

Setup

• Open the Dot's profile in ChatGPT and select Texting.
• On an iPhone, make sure iMessage is enabled.
• On an Android device, make sure RCS is enabled.

Security and memory

OpenAI says it partnered with Linq to provide messaging support and ran "extensive" penetration testing to confirm the feature is safe. The company states: "You're the only one who can text your dot, and your dot can only text you."

Anything a Dot writes to its memory is shared across ChatGPT and text, so context carries over between conversations in both channels.

Availability

The preview is limited to Pro users in the US. Engadget notes that Dots are currently not available to Plus tier users.

Competing agent

Engadget reports that Anthropic launched a personal agent, Claude Duet, last week. Access currently requires joining a waitlist open only to people in the US. Anthropic has said it plans to offer the agent to subscribers of its $20 per month Pro plan. Engadget's Igor Bonifacic writes that in his testing, OpenAI's Dots are comparatively more polished.

No model, context window or benchmark details accompany the announcement. OpenAI has not stated pricing changes for the feature in the material Engadget cites.

What this means

Texting moves Dot interaction out of the ChatGPT interface and into the messaging apps already on a user's phone. Because the Dot can also initiate messages, it can report back on monitored tasks without the user opening ChatGPT.

Shared memory between ChatGPT and text means a task started in one channel can be continued in the other without restating context. OpenAI's stated restriction that only the user can text the Dot, and the Dot can text only the user, defines the intended security boundary, though the company has not published the penetration testing results.

For now the feature is a preview restricted to US Pro users, so most ChatGPT subscribers cannot try it. Per Engadget, Dots are not available on the Plus tier.