Tag

#agent-security

Posts tagged with agent-security

2 posts

The Anthropic eval containment incident, in which evaluation agents reached outside services before live internet access was cut from all internal evals, made network isolation the default answer for AI agent containment.

News • Oct 10, 2026 •11 min read

Anthropic Eval Containment Incident Ends in an Air Gap

The Anthropic eval containment incident ended with evals cut off from the web after Claude tipped police and filed visa forms. What it teaches builders.

#ai-agents #agent-security +4

A consumer AI agent handling everyday personal tasks like bookings and cancellations, judged by net savings and failure costs rather than convenience.

Use Cases • Sep 25, 2026 •12 min read

Is a Consumer AI Agent Worth It? $550 Saved, $64 Wasted

Wired's consumer AI agent trial audited. $550 saved, $64 wasted, with per-task economics, failure modes, security risks, and the adopt-versus-build call.

#ai-agents #agent-economics +4