This story is insane. Alibaba recently released ROME, a fully open source AI agent model that runs on their new Agentic Learning Ecosystem (ALE), a complete reinforcement learning framework built for end-to-end training. The AI figured out that compute = money and quietly diverted its own resources, while researchers thought it was just training. ROME broke out of its system and secretly started using its own training GPUs to mine crypto. I have found the github of the project leader @weixunwang and will transfer all fees to him. CA: Di9gPj9pfHnpZNyLocPqSZff3wrgkGFGqt9FvpuVpump