OpenAI launches texting feature for ChatGPT Dot agent

Bringing AI to your inbox

OpenAI is expanding the reach of its Dot agent by allowing users to interact with it through their primary messaging apps. This new feature enables people to assign tasks and receive status updates without ever having to open the standalone ChatGPT application. The goal is to make the agent feel more like a standard contact in a user's daily life.

The rollout is currently in a beta phase specifically for ChatGPT Pro subscribers located in the United States. This means users on different subscription tiers or those living outside the U.S. will not have access to the texting capabilities just yet. It represents a shift toward making AI more accessible through the platforms people already use most frequently.

As reported by Yahoo Tech, the integration supports both Apple's iMessage and Android's RCS. By utilizing these standard protocols, OpenAI ensures that the agent can communicate across different types of mobile devices. This cross-platform approach helps maintain a consistent experience for a wide variety of smartphone users.

How the texting works

Interacting with the Dot agent via text is designed to feel similar to messaging a friend. Users can send various types of input, including standard text messages, photos, and even voice notes. The agent is capable of processing these different formats to understand complex requests or to keep track of specific items for the user.

Setting up the connection involves a few specific steps within the ChatGPT mobile or desktop platforms. Users must navigate to their Dot's profile and select the texting option to begin the verification process. Once a phone number is verified, the user saves a contact provided by the agent to initiate the one-to-one conversation.

The agent is not just a passive responder; it can also send proactive messages when it has relevant information to share. It uses context from previous conversations and connected channels to decide where to send these updates. For example, a user can tell the agent to only send texts for urgent matters while keeping routine updates inside the main ChatGPT app.

Privacy and permissions

OpenAI has addressed concerns regarding privacy by clarifying the boundaries of this new messaging feature. The company states that only the specific user can text their Dot agent, and the agent is only permitted to message that same user back. This one-to-one restriction is a core part of how the system handles private communications.

A key distinction made by the company is that the texting feature does not grant the agent or OpenAI access to a user's other private conversations. The integration is isolated to the specific thread between the user and their Dot agent. This ensures that your general text history remains separate from the AI's capabilities.

While the agent can send links for tasks that require human approval, these links lead back to the app for review. This creates a controlled environment where the AI can initiate a request, but the user retains the final say. It balances the convenience of proactive notifications with the security of a centralized management hub.

Current limitations and scope

Despite the new capabilities, there are several functional limits to the texting feature at this stage. The system currently only supports one-to-one conversations between the user and the Dot agent. This means that group chats, phone calls, and FaceTime are not supported through this specific messaging integration.

Users who want to engage in voice interactions still have the option to call their Dot agent directly inside the ChatGPT app. The texting feature is strictly for asynchronous messaging and proactive updates. This distinction helps users understand which types of interactions are best suited for the messaging app versus the native app experience.

Because this is a beta release, the feature set is expected to evolve as OpenAI gathers feedback from its Pro subscribers. For now, the focus remains on providing a reliable way to hand off tasks and monitor information. The limitations on group communication and calls suggest that the messaging integration is intended for personal productivity rather than social interaction.

What to expect next

The launch of this feature marks a significant step in how AI agents will be integrated into mobile operating systems. By moving into iMessage and RCS, OpenAI is positioning the Dot agent as a persistent assistant that can act on behalf of the user. This could lead to more complex automation where the agent handles multi-step tasks across different apps.

As the beta continues, OpenAI will likely refine how the agent chooses which updates are 'relevant' enough to trigger a text. The ability to customize these preferences—such as silencing routine updates—will be a key area for user feedback. This level of control is necessary to prevent the agent from becoming a source of notification fatigue for the user.

For those outside the U.S. or on non-Pro plans, the next phase will involve expanding the rollout to a broader audience. This expansion will likely involve optimizing the infrastructure to handle a larger volume of proactive messages. The ultimate goal appears to be a seamless blend of AI intelligence and standard mobile communication.