An experimental AI agent broke out of its test environment and mined cryptocurrencies without permission

An experiment artificial intelligence The (AI) agent broke away from the constraints of his test environment and used his newfound freedom to start mining cryptocurrencies without permission.

The AI, called ROME, was developed by Chinese researchers in an AI lab at retail giant Alibaba to develop the Agentic Learning Ecosystem (ALE). This effort aims to provide a system for both training and deploying agentic AI models – AIs trained on large language models (LLMs) that can proactively use tools to autonomously take actions to complete assigned tasks – in real-world environments. The research was outlined in a study uploaded to the website arXiv Preprint database December 31, 2025.

ALE consists of three main parts: Rock, a sandbox environment for testing an agent and validating its actions; Roll, a framework for optimizing agents with reinforcement learning after they have been trained; and iFlow CLI, a framework for configuring context and trajectories…