OpenAI’s take on AI agents could come in January

OpenAI could release an “AI agent” tool as soon as January.

That’s according to Bloomberg, which reports that OpenAI is close to launching software, code-named “Operator,” that can take direct actions on a person’s computer. OpenAI is said to be planning to initially launch Operator as a research preview through its developer API.

Operator would go up against Anthropic’s recently unveiled agent feature, Computer Use, and Google’s rumored consumer-focused agent, among others. It’s unclear whether OpenAI’s will have any advantage; Bloomberg reports that it could take the form of a general-purpose tool that executes tasks in a web browser.

The report on Operator arrived the same day as the release of OpenAI’s policy paper with suggestions for the U.S.’ AI strategy. That paper reportedly proposes, among other things, that the federal government create AI-focused “economic zones” permitting work to build AI infrastructure, and the establishment of a bloc with U.S. allies to compete with China on AI.