ChatGPT Dots Text US Pro Users Without End-to-End Encryption

OpenAI began rolling out texting for ChatGPT dots to US Pro users on October 28, 2026, according to its October 28 release notes. In this limited beta, an eligible user can give a dot work through iMessage or RCS and receive updates in the phone’s messaging app. The text thread reaches the same dot the user works with in ChatGPT, with its existing memory and connected plugins.

The convenience comes with a boundary for client work. Windows Report’s account of the rollout explains that Linq, a third-party messaging provider, processes texts on their way to OpenAI, so the conversation is not end-to-end encrypted between the user and the dot. A message sent from a familiar phone app therefore passes through another provider before the dot receives it.

What a text can set in motion

Texting gives an existing dot another way to receive instructions, rather than creating a separate agent for the phone. A user can hand off a task, add context to work already in progress or ask the dot to keep an eye on something. The channel accepts text, photos and voice messages, so a handoff can be more than a typed command. What the dot can do with it still depends on the apps and permissions connected to that dot.

The dot can also initiate an update when it has something relevant to report. It uses the conversation’s context to choose where to reach the user, who can ask for particular updates by text while leaving routine progress in ChatGPT. For an online professional, that could mean asking for a short alert when a monitored task needs a decision, then reviewing the underlying work in ChatGPT instead of trying to manage a long project through message bubbles.

An approval request has its own handoff. If the dot needs approval, it sends a link to review the proposed action in ChatGPT. The text is a notification that a decision is waiting; the review takes place in ChatGPT. That distinction matters when a short instruction could lead to an action in a connected app.

Who can connect a phone

The texting audience is narrower than the audience for dots. The Daily Bot’s rollout brief records the limited US Pro beta, the exclusion of Business and Enterprise workspaces, and the absence of SMS and group-text support. A Pro subscription alone does not mean the option has reached an account. An eligible user connects a phone through the dot’s Texting setup and completes phone-number verification.

On an iPhone, the route is iMessage; on Android, it is RCS. The conversation is one-to-one: the account holder texts their own dot, and the dot texts that person. A group thread cannot bring coworkers or a client into the exchange. Texting also does not give the dot or OpenAI access to the user’s other phone conversations. Those limits describe the messaging channel, while any apps already connected to the dot remain a separate source of context for its work.

The workspace exclusion is especially relevant to a professional whose client material sits in a managed Business or Enterprise account. Availability in a personal Pro account does not extend this beta to that work account. A phone conversation with a personal dot should be judged by the permissions and information available to that dot, rather than by the controls a different workspace may provide.

Where client material crosses the boundary

Unpublished interview material stays separate from a short instruction sent through ChatGPT dot texting.

The privacy issue concerns the full route from the user to the dot. Linq’s role in delivering and processing messages means the iMessage or RCS label on a phone does not establish end-to-end encryption for the complete dot conversation. This is a material distinction for an unpublished transcript, a confidential brief, customer details or a credential copied into a text. Each would become part of the message sent through that route.

There is also a boundary inside the dot’s ongoing work. The texting channel carries over its memory and connected plugins, so a brief instruction may let the dot use context the user previously supplied or an app the user previously connected. Sending a text does not itself connect a new inbox or project. Conversely, keeping a sensitive passage out of the text does not settle whether the dot already has permission to read that passage through another connection.

For example, a creator could ask a dot to prepare a summary from an interview already held in a connected project, provided they have permission to use that project with the dot. That avoids pasting the transcript into the phone thread, but the dot’s access to the project still matters. Copying the entire unpublished interview into a text makes a different disclosure decision: the source material itself travels through the messaging provider.

The decision matrix for a work handoff

The beta’s useful boundary is whether a text contains a short instruction or the sensitive material behind the task. The following distinctions bring its access, channel and privacy rules together for creator work:

Eligibility: A selected US Pro user can connect a phone when Texting appears for their dot. Business and Enterprise workspaces are outside this texting beta.
Supported channel: An individual iMessage or RCS exchange can carry text, a photo or a voice message. SMS and group threads are unsupported, so the dot cannot join a client team’s group conversation.
Proactive updates: The dot can send a relevant update without a fresh text from the user. A request for a brief alert about an assigned task fits that channel better than moving the entire working record into it.
Approval links: A text can point to an action awaiting review. The user follows the link into ChatGPT to inspect the request.
Message privacy: Linq processes the messages, and the user-to-dot conversation lacks end-to-end encryption. Confidential source material and credentials are poor candidates for the text thread.
Safer creator handoff: A short request to work from an already authorized project avoids reproducing its contents in a phone message. The project connection and the dot’s permissions still need to suit the assignment.
Riskier creator handoff: Pasting a private client brief, customer list or unpublished transcript puts that material directly into the third-party messaging route.

For eligible users, the immediate gain is the ability to direct an ongoing dot and receive timely prompts away from ChatGPT. The delivery path sets the limit on what belongs in those prompts: a concise task or update can travel by text without making the text thread the place where a client’s underlying material lives.