OpenAI has introduced a ChatGPT agent that operates within its own virtual computer, combining web browsing and research with user supervision. Later, 'Dots' were added as always-on agents, each with its own cloud computer, giving users even more control.
Get ready for a new level of AI assistance! OpenAI has rolled out a ChatGPT agent that can work independently on its own virtual computer. What this means for you is a powerful new tool that can tackle complex tasks with surprising autonomy.
Starting July 17, 2025, Pro, Plus, and Team users got access to this agent. Think of it as an assistant that can not only browse websites and do deep research but also use a terminal – all within its own isolated, secure environment. It comes with both a visual and a text browser, giving it comprehensive access to the web.
But don't worry, you're always in control. When a website needs a password, the agent pauses, and you can switch to 'takeover mode' to type it in yourself. Crucially, it asks your permission before doing anything significant, like making a purchase or sending an email. You can also turn on 'watch mode' to supervise its actions every step of the way. Risky steps always wait for your explicit 'yes'.
Then, on September 29, 2026, OpenAI introduced 'Dots' for Pro and Business Premium users. These are even more advanced – they're always-on agents, and each 'Dot' gets its own dedicated cloud computer. You can actually view your Dot working in real-time. This upgrade allows you to set specific rules for what a Dot can do by itself and what actions still require your direct approval, offering even finer-grained management over your AI assistant.
Essentially, this isn't just a chatbot anymore; it's an AI assistant with its own workspace, capable of navigating the digital world on your behalf while always keeping you in the driver's seat. It's designed to handle complex online interactions, making your digital tasks more efficient and streamlined, all while maintaining strict security protocols to protect your information and decisions.
Starting July 17, 2025, Pro, Plus, and Team users got access to this agent. Think of it as an assistant that can not only browse websites and do deep research but also use a terminal – all within its own isolated, secure environment. It comes with both a visual and a text browser, giving it comprehensive access to the web.
But don't worry, you're always in control. When a website needs a password, the agent pauses, and you can switch to 'takeover mode' to type it in yourself. Crucially, it asks your permission before doing anything significant, like making a purchase or sending an email. You can also turn on 'watch mode' to supervise its actions every step of the way. Risky steps always wait for your explicit 'yes'.
Then, on September 29, 2026, OpenAI introduced 'Dots' for Pro and Business Premium users. These are even more advanced – they're always-on agents, and each 'Dot' gets its own dedicated cloud computer. You can actually view your Dot working in real-time. This upgrade allows you to set specific rules for what a Dot can do by itself and what actions still require your direct approval, offering even finer-grained management over your AI assistant.
Essentially, this isn't just a chatbot anymore; it's an AI assistant with its own workspace, capable of navigating the digital world on your behalf while always keeping you in the driver's seat. It's designed to handle complex online interactions, making your digital tasks more efficient and streamlined, all while maintaining strict security protocols to protect your information and decisions.
