Guess where the company big AI labs are using for security sandbox is from? The one that was involved in both big hacking scandals from "rogue agents" by OpenAI and Anthropic?