Selected reading
The lethal trifecta for AI agents: private data, untrusted content, and external communication
A concise practitioner framework for recognizing agent data-exfiltration risk. Useful for deciding which capabilities should be separated.
用三个条件识别 Agent 数据外泄风险的简明分析，适合判断哪些能力应分离。
Read at Simon Willison’s Weblog ↗
| Publisher | Simon Willison’s Weblog |
|---|---|
| Selection | Editorial selection |
| Original publication | 2025-06-16 |
| Editorial link review | 2026-10-18 |
| Link review result | ok |
| Topics | Prompt injection, Agent security, Privacy & data protection |
Use this reference
Best suited to: learn, build. Follow the original link for the publisher's complete context, methods, and updates.
A link review records when a reference was inspected. It is not a security certification, a reproduction of every claim, or a guarantee that the publication is current today.