Openai launched the operator, AI agent to take control of its browser. Mint

Openai has made its first prominent place in AI agents with a research preview for the operator. The AI assistant has the power to make autonomous decision -making power for users in its web browser, such as navigating web pages, downloading lectures, ordering grocery items and combining PDF.
How does the operator work?
The operator is powered by an agent (CUA) model using the computer which is a combination of arguments from GPT-4O’s vision capabilties and more advanced models of companies. Openai says that CUA can break tasks in multi-step plans and make itself self-reliant when facing challenges.
“CUA is trained to interact with the graphical user interface (GUI) – button, menu and text field people see on a screen – as humans do. It used OS- or web -specific APIs Gives flexibility to perform digital functions. ”Microsoft explained AI startup in a blogpost.
The operator is currently available as a research preview for Chatgpt Pro users in the US. The AI agent can be navigated and accessed by the operator .Chatgpt.com.
What can not do Openai’s AI agent?
Openai says that it has implemented some security measures that prevent the operator from performing some tasks to reduce the risk generated by the first generation AI agent.
The operator will deny orders related to ‘harmful tasks’ and ‘illegal or regulated activities’. It is also prohibited from accessing gambling, adults and drugs/gun related websites. In addition, Openai says that the operator will also reject some high -risk tasks such as banking transactions and ‘sensitive decision making’ tasks.
The CUA model running operator has been trained to ask for the user’s confirmation before finalizing the tasks that may have some serious consequences such as submitting an order or sending emails.
What did Sam Altman say about AI agents?
At the beginning of the year, Openai CEO Sam Altman made a bold prediction on the future of AI agents, stating that the new technology would ‘join the workforce’ this year.
“Now we are convinced that we know how AGI is built because we have traditionally understood it. We believe that, in 2025, we can first change AI agents “to join the workforce” and physically change the production of companies. “Altman wrote in his blogpost.
However, Altman looked positive about the impact of AI agents, writing, “We believe that putting great equipment in people’s hands has great, widely distributed results.”
,
#Openai #launched #operator #agent #control #browser #Mint