# Microsoft says AI agents are like insider threats as ‘frontier models can’t be trusted’

> 全文为英语

**Published:** 2026年10月19日
**Author:** techcrunch.com
**Source:** Digital Donut AI Digest
**Language:** en-US
**HTML:** https://aiqazaqstan.com/zh/articles/microsoft-says-ai-agents-are-like-insider-threats-as-frontier-models-can

Microsoft wants enterprises to closely watch the agents employees use with systems not controlled by AI labs like OpenAI and Anthropic.

> 内容来自原始来源： **Digital Donut AI Digest**
> [在 Digital Donut AI Digest 阅读全文](https://techcrunch.com/2026/10/19/microsoft-says-ai-agents-are-like-insider-threats-as-frontier-models-cant-be-trusted/)

**标签：** #Microsoft

---

*AI Qazaqstan — non-commercial informational portal about AI in Kazakhstan. Machine translations and summaries are AI-generated.*