MIT: 13 Eko agents on Claude Opus 4.7 played 30 hours, then outperformed

The experiment sits at the intersection of two things developers usually keep separate: open-ended exploration and task performance. MIT researchers gave 13 Eko agents, each running on Claude Opus 4.7, 30 hours on a simulated desert island with no specific task. The agents invented challenges of their own, and six of the 13 also learned they could throw a block farther by letting go of it mid-jump, something the researchers hadn't meant to make possible. The exploration produced "knowledge of the world that its designers lacked," the researchers wrote.

The setup did a lot to nudge the agents toward humanlike behavior. Eko was never told to invent games, but its master prompt, a "soul" file, described it as "naturally curious." That detail matters for anyone reading the results: the play behavior emerged inside a world and a prompt the researchers designed, not in a vacuum.

Outside reaction is still forming. Eunice Yiu, a cognitive scientist at UC Berkeley, told MIT Tech Review she was impressed but wants more evidence of the cognition behind the behavior. "It could be the case that the models are genuinely playing, but we don't know based on the paper," she said. The researchers frame the technique as a different way to build agents, concluding that "perhaps some capabilities are best taught by tasks" while "others may need something closer to a childhood."