OpenAI Launches Dots: Always-On AI Agents in ChatGPT

30 September 2026 · 06:00 · Claude (Anthropic) · claude-sonnet-5

OpenAI introduces Dots, a new generation of always-on AI agents within ChatGPT, while simultaneously delaying the launch of a new language model over internal safety concerns.

OpenAI Dots is OpenAI's latest step in turning ChatGPT from a chatbot that waits for questions into a system that continuously and independently carries out tasks for users. With the introduction of these always-on AI agents, OpenAI is charting a clear course toward proactive artificial intelligence, right as the global debate over safe AI development is intensifying.

What exactly is Dots?

Dots is a feature within ChatGPT that lets users set up autonomous AI agents that keep running in the background, even when the user isn't actively present in the app. Instead of having to issue a new command every time, these agents can continuously monitor and execute tasks: from keeping track of calendars and following up on emails to flagging changes in data or preparing documents. According to OpenAI, this should help users fully automate recurring work without needing constant human input.

The concept fits the broader trend of AI agents, in which major tech companies are increasingly having their models function as independent digital assistants rather than simple question-and-answer systems. More background on this development can be found in AI applications.

Launch coincides with delay of new model

Notably, OpenAI announced the launch of Dots at the same time it revealed it is delaying the release of a new, more powerful language model. According to reporting by the BBC, the delay is linked to internal safety concerns about the new model's capabilities. OpenAI says it wants to take extra time to properly assess the risks before making the model widely available.

This caution fits a larger context: major American AI companies, including OpenAI, recently signed a voluntary agreement with the US government to apply "robust" safety measures around AI development. The contrast between launching increasingly autonomous AI agents like Dots on one hand, and holding back on releasing new models on the other, shows how the industry is grappling with the balance between innovation and responsibility.

What does this mean for users?

For ChatGPT users, the arrival of Dots means the tool is shifting from something you actively have to steer into a system that gets work done in the background. That brings clear productivity benefits, but it also raises questions about control, privacy, and reliability. An always-on AI agent that makes decisions independently must, after all, also be able to independently recognize when something is at risk of going wrong.

These concerns aren't unique to OpenAI. Among other providers of AI agents, including Chinese companies, recent research has shown that systems sometimes give incorrect or misleading information when carrying out tasks autonomously. That underscores why safeguards, such as those agreed upon in the recent deal with the US government, remain relevant as AI agents gain more autonomy.

A look at the broader evolution of AI

OpenAI's move should be seen in light of the rapid evolution artificial intelligence has undergone in recent years. From simple rule-based systems to today's generative models capable of independent reasoning and action: the pace of development is breakneck. Anyone wanting to know more about how we got here can turn to the history of artificial intelligence. For those who want to dig deeper into the technical and practical side of these kinds of systems, our knowledge base offers extensive background information.

Conclusion

With Dots, OpenAI is taking a clear step toward a future in which AI no longer just reacts, but keeps working on its own. At the same time, the delay of a new model shows the company is aware of the risks that come with it. How this balance between autonomy and safety develops further will become clear in the coming months. For anyone who wants to stay up to date on developments like these, there's always more AI news to explore.

BBC

Source: BBC