OpenAI’s operator can browse the web for you

OpenAI has started previewing a new tool called Operator that can navigate in a web browser. According to a blog post published on Thursdaythe software is powered by what the company calls a Computer-Using Agent. “CUA is designed to interact with graphical user interfaces (GUIs)—the buttons, menus, and text fields that people see on a screen—like humans,” OpenAI says of the model. “This gives them the flexibility to perform digital tasks without using specific OS or Web APIs.”

The current version of Operator is based on OpenAI’s GPT-4o model. It combines the vision capabilities of that algorithm with “advanced reasoning” trained for reinforcement learning. The operator has the ability to “break tasks into multi-step plans and adaptively self-correct when challenges arise.” According to OpenAI, this capability represents the next stage in the development of AI.

As with past research previews, OpenAI warns that Operator is “still early and has limitations,” and that it “does not have reliable performance in all scenarios yet.” For example, depending on the complexity of the task and the interface involved, the agent will benefit greatly from the user taking a few extra moments to write a more detailed prompt. In order The VirginThe operator will give the user control if they are still stuck in a task. It will also send the check whenever a website asks for sensitive information, including login credentials. The company says it designed the tool to “reject malicious requests and block unauthorized content.”

OpenAI makes Operator first available to users of its $200 per month ChatGPT Pro Subscription. It also partnered with companies like Instacart to offer the agent on their platforms, although a ChatGPT Pro subscription is also required to test the integration.

The operator joins a growing list of AI agents that can navigate a web browser or an entire operating system. Anthropic was the first to offer the capability with the release of its Claude 3.5 Sonnet model in Octoberfollowed most recently by Google with its Gemini 2.0 model and Project Mariner.

If you buy something through a link in this article, we may earn a commission.