I'd argue that a lot of developers can't determine if an LLM generated command is actually safe or not.