OpenAI has announced the Operator, an AI agent that can navigate the web and perform a variety of tasks. The agent is currently available to ChatGPT Pro users in the US for $200 per month, with plans to expand to Plus, Team, and Enterprise users in the future.
Operator can automatically handle tasks that you use on the web. It can perform a variety of tasks, such as canceling orders, calculating refunds, searching for customers in your internal database, writing emails, and purchasing groceries. On your computer, it can also perform advanced tasks, such as downloading files, merging PDFs, analyzing spreadsheets, and exporting images.
The core model of Operator is a 'Computer-Used Agent (CUA)', which uses GPT-4o's vision mode to understand the content through screen shots and manipulate the graphical user interface. This allows it to perform actions on the web browser, such as clicking buttons, typing, and scrolling.
OpenAI has implemented various measures to ensure the safety of the Operator. The Operator blocks illegal or harmful actions and does not access blacklisted sites such as gambling, adult content, and weapons sales sites. The automated safety check system monitors user actions in real time and can issue warnings or blocking in case of policy violations.
Costly or sensitive tasks require user verification, and high-risk tasks such as financial transactions are restricted. Through these measures, OpenAI focuses on ensuring that users trust the Operator and reducing the potential risks of AI.
With this announcement, OpenAI is one step closer to its goal of making 2025 the “Year of Agent AI.” Operator is gaining attention as an innovative technology that goes beyond simple automation tools to help users free up more time from repetitive tasks.
