OpenAI's AI agents now put in 3 days of work for every 1 day of human labor in the company's research org. And the thing those agents are working on is the next version of themselves. AI researchers have a name for that loop, recursive self-improvement. For decades it was a thought experiment in safety papers. This morning OpenAI published it as a set of internal charts. Last fall they set a public goal of an automated research intern by September 2026. Today they announced they hit it. A…