OpenAI Letting some users try out a new ChatGPT feature that uses artificial intelligence Operating a web browser to book trips, buy groceries, shop for bargains, and many other online tasks.
The new tool, called Operator, is an AI agent: It relies on an AI model trained on both text and images to interpret commands and learn how to use a web browser to execute them. OpenAI claims that it has the ability to automate many daily tasks and workday chores.
OpenAI's operator follows rival releases from both Google and anthropic, which is displayed Able to access the web. AI agents are It is widely seen as the next evolutionary stage AI is following chatbots, and many companies have jumped on the hype train by promoting them. In most cases, these are very limited in their capabilities and simply use a language model to automate things normally done with regular software.
“AI is evolving from a tool that can answer your questions to one that is also capable of taking action in the world by carrying out complex, multistep workflows,” says Peter Welinder, VP of Product at OpenAI. “We will see a huge impact on people's productivity—but also on the quality of work that people are able to complete.”
OpenAI acknowledges that giving ChatGPT access to web browsers creates new risks, and says operators may sometimes misbehave. It says it has implemented a number of new security measures and plans to gradually enhance operator capabilities.
Velinder and Yash Kumar, head of product and engineering for OpenAI's Computer Using Agent, say the plan is to learn how people use the tool. They acknowledge that the tool may lead to unwanted bookings or purchases, but they also say that a lot of work has been done to make sure it asks before doing anything risky. “It will come back to me and ask for confirmation before taking steps that may be irreversible,” says Kumar.
OpenAI also released a new “system card” today that outlines the issues the operator is facing. These include the possibility of commands being misunderstood or being different from what the user asked; to be misused by users; Or to be targeted by cyber criminals.
“It also presents an incredible amount of security challenges,” says Kumar, “because your attack vector area and your risk vector area increases significantly.”
The operator will initially be available as a “research preview” to ChatGPT users with a Pro account, which will cost $200 per month. The company says it plans to expand reach by releasing the tool gradually, as it will inevitably make some mistakes along the way.
In several demonstrations, operators showed the potential for AI to play a more active role as a web assistant. The tool features a remote web browser and a chat window to communicate with the user.
At WIRED's request, the operator was asked to book Amtrak train travel from New Haven, Connecticut to Washington, DC. It went to the correct website and entered the required information correctly to bring up the timetable, then asked for further instructions. If a user was logged into an Amtrak website or browser profile with stored credit card information, the operator would be able to go ahead and book a ticket – although it is designed to ask for permission first.
Kumar asked the operator to book a table at Beretta, a restaurant in San Francisco. The program went to the OpenTable website, found the right restaurant, and checked availability before asking what to do next. OpenAI says it has partnered with many popular sites, including OpenTable, to ensure the operator works smoothly on them.
The new tool is based on OpenAI's GPT-4o AI model, which can look at browsers and web pages and interact with typed text. The tool includes additional training designed to help you understand how to perform tasks online. OpenAI will also make its computer usage agent available through its API.