MCP protocol found vulnerable to cross-agent prompt injection
What happened
From the reports
独立研究者 Syed Anas Mohiuddin 测试了 Google、摩根大通、Weaviate、Rapid7、法国政府部际数字事务局和美国联邦政府等机构的智能体，利用 MCP（Model Context Protocol）中的信任缺口实施概念验证攻击。
From Ars Technica · AI
Timeline
Follow the reports to see the event from each side.
Oct 6
- Ars Technica · AIMCP for agent-to-agent comms may be the riskiest protocol you've never heard of
Trust gaps in the new protocol spread malicious prompts from one agent to another.