OpenAI could release an “AI agent” tool as early as January.
This is according to Bloomberg, which reported that OpenAI is close to launching software code-named Operator that can perform actions directly on a person’s computer. OpenAI is said to be initially planning to launch Operator as a research preview via its developer API.
Operator will go up against the likes of Anthropic’s recently announced agent feature, Computer Use, and Google’s rumored consumer-focused agent. It’s unclear whether OpenAI has any advantages. Bloomberg reports that it could take the form of a general-purpose tool that performs tasks in a web browser.
The report on Operator arrived on the same day as the release of OpenAI’s policy document, which includes implications for U.S. AI strategy. The paper calls for, among other things, creating an AI-focused “economic bloc” in which the federal government would authorize work to build AI infrastructure, as well as establishing a bloc with U.S. allies to compete with China on AI. He is reportedly proposing that.
