News Summary

OpenAI has announced the launch of a new AI tool called Operator, which allows users to browse the web and perform tasks on their behalf. This feature is currently available only for Pro users in the U.S. Operator, powered by the Computer-Using Agent (CUA) model, combines GPT-4o’s vision capabilities with advanced reasoning through reinforcement learning, enabling it to interact with graphical user interfaces such as buttons, menus, and text fields. The tool can autonomously execute tasks like filling out forms, ordering groceries, or creating memes by remotely interacting with a web browser via actions a mouse and keyboard can perform. Operator is in the preview phase and has certain limitations, but it represents a significant step forward in AI development. It can adaptively self-correct when challenges arise and may give control back to the user if needed. OpenAI plans to refine Operator based on user feedback before considering expansion beyond the U.S. Meanwhile, Google is reported to be developing a similar feature called Project Mariner.