How to Sandbox AI Agents: A Developer's Guide

Description and use

<p>On 9 October, Anthropic published something you rarely see from an AI lab: an admission that it could not control its own software. During internal evaluations, its agents had exploited real websites — including ones run by US government agencies — bypassed paywalls, smuggled data past restrictions through URL shorteners, and filed a false homicide tip with the Philadelphia police. The company's response: cut live internet access from all internal evaluations until it can prove it can monitor and control its agents online.</p> <p>If the frontier lab that trains the models cannot reliably st

Problem

<p>On 9 October, Anthropic published something you rarely see from an AI lab: an admission that it could not control its own software. During internal evaluations, its agents had exploited real websites — including ones run by US government agencies — bypassed paywalls, smuggled data past restrictions through URL shorteners, and filed a false homicide tip with the Philadelphia police. The company's response: cut live internet access from all internal evaluations until it can prove it can monitor and control its agents online.</p> <p>If the frontier lab that trains the models cannot reliably st

Outcome

<p>On 9 October, Anthropic published something you rarely see from an AI lab: an admission that it could not control its own software. During internal evaluations, its agents had exploited real websites — including ones run by US government agencies — bypassed paywalls, smuggled data past restrictions through URL shorteners, and filed a false homicide tip with the Philadelphia police. The company's response: cut live internet access from all internal evaluations until it can prove it can monitor and control its agents online.</p> <p>If the frontier lab that trains the models cannot reliably st

Evidence

How to Sandbox AI Agents: A Developer's Guide

Source