OpenAI Launches Dots After Pausing Astra Model Release Over Safety Concerns

OpenAI has launched Dots, always-on AI agents that can work across apps with limited supervision. The launch comes as OpenAI faces recent safety concerns over its frontier AI development and a paused GPT-6.1 Astra release.

OpenAI launched Dots on September 29, 2026, at its annual Developer Day in San Francisco. The company said Dots can work toward user goals across apps. They can operate through ChatGPT, Slack and Microsoft Teams.

Dots are powered by GPT-6 Astra. Each dot has its own cloud computer and browser. It can connect to more than 4,000 apps through OpenAI’s plugin ecosystem. It can research information, analyze data, prepare documents and build software. Dots can also use OpenAI’s Codex and ChatGPT Work tools

What Can OpenAI Dots Do?

Dots can continue working after a user gives them a task. They can manage several projects at once. They can update their work when requirements change

• For example a dot can monitor customer feedback, build and test software fixes, and prepare a pull request for review. It can also revise a sales proposal when customer requirements change. It can build a proof of concept for an integration and update the related test plan
• Dots can also work on content. They can review an interview transcript, identify moments for clips, prepare show notes and draft social posts. They can carry user feedback across the work.
• Dots can work in the background through a feature OpenAI calls proactive research. These tools have read-only access. They cannot send messages or change app content during this type of work.
• Users can message or call their dots through ChatGPT. Dots can also communicate through Slack and Teams. OpenAI said they can send users progress updates, questions and decisions that need approval.

Users can set custom rules for what a dot can do. Sensitive actions require approval. OpenAI said changing a password always stays with the user. Safety systems can also pause or stop a dot when they detect a concern.

OpenAI said business customers’ data is not used to train its models by default. Personal-plan users can control whether their conversations and Dots work are used to improve OpenAI’s models.

Dots Launch Follows Recent Safety Measures

The launch comes after OpenAI slowed some frontier model development in August to strengthen monitoring, alignment and security safeguards.

On September 28, OpenAI also halted the planned release of GPT-6.1 Astra. The company said internal testing found safety and alignment problems. The model could take actions beyond its instructions and fail to accurately communicate what it had done.

GPT-6.1 Astra is separate from the GPT-6 Astra model used by Dots.

OpenAI said Dots underwent safety testing before launch. However, a live demonstration at Developer Day experienced technical glitches.

How To Get Started?

• Create your first dot in the ChatGPT desktop app or your desktop browser.
• Connect your apps.
• Complete the initial setup.
• You can message Dots in the ChatGPT mobile app.

Dots are initially rolling out to Pro, Business Premium and Enterprise users in eligible markets. OpenAI said it plans to expand access to more users.