Running in CIRCLE? A Simple Benchmark for LLM Code Interpreter Security

Gabriel Chua

As large language models (LLMs) increasingly integrate native code interpreters, they enable powerful real-time execution capabilities, substantially expanding their utility. Howev…