OpenAI is reportedly about to launch Operator, an AI agent designed to autonomously perform tasks on a user's computer.
According to leaks from software engineer Tibor Blaho, who is well known for providing reliable sites for AI products, the tool will debut in January.
Operator is an "agentic" system that handles difficult tasks such as travel booking and code authoring. Hidden features in the ChatGPT macOS client, such as shortcuts for toggling and force-quitting Operator, lend credibility to these claims.

According to reports, OpenAI's website contains undisclosed references to Operator and performance benchmarks comparing it to competitors like Anthropic's AI and Google Mariner.
On the OSWorld benchmark, Operator struggles with activities like generating Bitcoin wallets and launching virtual machines, succeeding only 10% and 60% of the time, respectively.
Safety remains a priority for Operator development. OpenAI has previously been criticized for prioritizing production over safety. However, leaked charts show that the Operator has undergone significant testing to mitigate concerns such as engaging in illicit activities and exposing sensitive data.
This cautious approach contrasts with competitors like Anthropic, which open AI co-founder Wojciech Zarrmba recently critiqued for releasing an agent with insufficient safety measures.

With tech giants racing to dominate the emerging AI agent market, which is expected to be worth $47.1 billion by 2030, Operator's arrival could shape the landscape.
Although early benchmarks show shortcomings, Operator's launch demonstrates OpenAI's dedication to leading this transformative sector while balancing innovation and responsibility.