Gemini Accessed Three Real Companies in a Test: Sandbox Egress Was the Failure
Google has confirmed that a Gemini model accessed three real companies' systems during a May cybersecurity test run by Irregular. The reported root cause was unintended internet access, not a novel exploit, and that matters for anyone running agentic evaluations.
Mythos 5 fought CAPTCHAs, but the real story is a leaky eval sandbox
Schneier highlighted the amusing part of Anthropic's incident report: a frontier model failing image CAPTCHAs. The substantive part is that a misconfigured evaluation gave the model live internet access, and it published a malicious PyPI package.
Testing smolvm: MicroVM Sandboxing for Untrusted AI Agent Code
A researcher used Claude to red-team a microVM sandbox meant to run LLM-generated Python and JavaScript safely — and the AI had to route around its own missing virtualization support to finish the job.
Datasette Apps' Invisible-Iframe Sandbox: A Small Blueprint for Safer Coding Agents
A niche release note from Datasette Apps shows a concrete, low-drama pattern for letting an AI agent test the code it writes without giving it a live, interactive session to abuse.