OpenAI has introduced an exciting new feature to its popular chatbot, ChatGPT, allowing users to assign tasks to the AI assistant with unprecedented autonomy. This development marks a significant step in the industry-wide push towards creating comprehensive digital helpers that can manage various aspects of our online lives.
ChatGPT’s agent mode, rolling out immediately, enables the model to execute actions on behalf of the user, utilizing its own virtual computer to handle complex requests. For example, users can instruct the AI to “look at my calendar and brief me on upcoming client meetings based on recent news” or “plan and buy ingredients for a Japanese breakfast for four.”
This feature is now available to users subscribed to Pro, Plus, or Team plans. It builds upon capabilities from existing tools like ChatGPT Operator and Deep Research, allowing the AI to browse the web and analyze online resources.
CEO Sam Altman acknowledges that this new functionality presents risks and emphasizes the importance of caution when granting access to personal information. He advises users to be mindful of what they share with ChatGPT, as the AI’s capabilities are still evolving.
The agent mode is part of a broader effort by tech giants to develop advanced digital assistants, rivaling Google’s Gemini helper. Apple is also working on an enhanced version of Siri that can perform tasks on behalf of users, although this update is delayed indefinitely.
As the industry continues to refine its AI agents, it’s essential to address concerns around hallucinations, bias, and unpredictable behavior. OpenAI is taking steps to mitigate these risks by limiting data access and requiring user oversight for certain tasks, such as sending emails or making bank transfers.
Source: https://edition.cnn.com/2025/07/17/tech/chatgpt-agent-think-and-act-for-you-new-update