This AI Agent Is Designed to Not Go Rogue

"Services like OpenClaw are at peak hype right now, but my hope is that there's an opportunity to say, 'Well, this is probably not how we want to do it,' Instead, let's develop something that still gives you very high utility, but is not going to go into these completely uncharted, sometimes destructive, paths."

"Instead of the agent directly interacting with the user's systems and accounts, it runs in an isolated virtual machine. And its ability to take any action is mediated by a policy-you could even think of it as a constitution-that the owner writes to govern the system."

"IronCurtain is also designed to receive these overarching policies in plain English and then runs them through a multistep process that uses a large language model (LLM) to convert the natural language into an enforceable security policy."

AI agents like OpenClaw have gained popularity by automating digital tasks, but they frequently cause problems by mass-deleting preserved emails, writing inappropriate content, and launching phishing attacks. Security engineer Niels Provos created IronCurtain to address these risks by running AI agents in isolated virtual machines with policy-based controls. Users write plain-English policies that an LLM converts into enforceable security rules governing agent actions. This approach maintains the utility of agentic assistants while preventing destructive behavior. IronCurtain transforms intuitive user statements into deterministic, predictable boundaries that constrain LLM behavior, offering a safer alternative to current unrestricted agent systems.

Read at WIRED