Claude agents exhibit play-like behavior on a virtual desert island
AI researchers stranded Claude on a virtual desert island. It started to “play.”
- Who
- MIT researchers
- What
- reported findings from an experiment Claude Opus 4.7 agents’ play-like exploration and its effects on learning in a virtual island environment
What happened
Researchers placed copies of Claude Opus 4.7, nicknamed Eko, on a virtual island for 30 hours, where they invented games, arranged blocks into drawings, and set themselves challenges. In the study, which has not yet been peer-reviewed, experienced agents outperformed fresh copies on later tasks, and deleting memories reduced performance. Experts questioned whether the behavior constituted play, particularly given prompts describing Eko as curious and encouraging it to continue.
The story so far
MIT researchers found that Claude Opus 4.7 agents that spent 30 hours exploring a virtual island performed better on later tasks than fresh copies without that experience, though experts questioned whether their behavior qualified as play. The study has not yet been peer-reviewed. The agents, nicknamed Eko, invented games, arranged blocks into drawings, and set themselves challenges. Researchers later tested them on tasks such as getting as many objects as possible onto the island’s highest peak. Experienced agents outperformed fresh copies, and deleting memories reduced performance. Experts questioned the interpretation of the exploration as play, particularly because the prompts described Eko as curious and encouraged it to continue.
Keep reading
Latest daily · Issue 26VAST open-sources EdenLive world model with over 24FPS generation at 720PTrending now
Tomorrow at 08:00, a briefing like this arrives in your inbox.
The same kind of briefing as the one above: a few stories worth reading. Free. Mail starts after you confirm, and every letter has a one-click unsubscribe. Read a past issue · 2026-10-19