OpenAI Announces ChatGPT Dots: Meet the New Persistent AI Agents

Introduction to ChatGPT Dots

For many ChatGPT users, the AI is primarily used as an interactive tool inside a chat window, requiring users to type a prompt, wait for an answer, and then manually transfer that response elsewhere. OpenAI is fundamentally looking to shift that paradigm with the introduction of ChatGPT Dots.

Announced today, Dots are persistent AI agents equipped with their own virtual cloud computers. This setup allows them to seamlessly browse the web, interact with connected applications, and perform multi-step tasks continuously—even after you have closed your browser tab or walked away from your device.

How ChatGPT Dots Work

Rather than simply asking ChatGPT to perform a task and leaving you to finish the execution, you can hand a complete assignment directly to a Dot. The agent operates within an isolated virtual machine in the cloud, keeping your personal computer safe and your local files secure by default.

• Isolated Cloud Environment: Dots execute workflows on a dedicated virtual machine rather than taking over your personal machine or crowding your local browser with endless tabs.
• App Integrations: Dots can connect to over 4,000 different third-party apps and services via plugins, extending their utility far beyond the standard chat interface.
• Cross-Platform Accessibility: You are not locked into the ChatGPT web interface. Dots can be accessed and communicated with through mobile devices, voice calls, Slack, Microsoft Teams, and upcoming SMS support.

Practical Use Cases for AI Agents

The true power of Dots shines through in administrative and everyday workflows that usually demand switching between multiple websites and applications. For instance, instead of merely asking ChatGPT where to order dinner, you can instruct a Dot to compare various local delivery menus, itemize cart totals including local taxes and tip estimates, and present a concise breakdown for your final approval.

In a professional setting, a Dot can ingest a video transcript or interview recording, help generate social media captions, pinpoint highlights, and organize asset files across connected platforms without requiring constant human oversight.

Privacy, Security, and Safety Measures

Giving an AI agent persistent capabilities and access to your accounts naturally raises security and privacy questions. OpenAI has built several vital safety constraints into the core architecture of Dots:

• Read-Only Background Tasks: While operating autonomously in the background, Dots are restricted to a read-only mode for information gathering. They cannot independently send messages, delete files, or complete financial transactions without explicit user authorization.
• Credential Separation: Dots can log into supported third-party services using safely stored credentials without ever exposing your raw passwords to the underlying AI model.

Availability and Pricing

ChatGPT Dots are rolling out to ChatGPT Pro and Business Premium subscribers in eligible regions, alongside enterprise-level beta access for administrators. For Pro users, the first Dot is included at no additional charge with their subscription.