# OpenAI 暂停 Astra 部分内部工作，称其网络安全能力或已达"严重"阈值

- 来源：Chubby♨️ (@kimmonismus)
- 发布时间：2026-08-08 04:17
- AIHOT 分数：64
- AIHOT 链接：https://aihot.virxact.com/items/cmsjeuusg07u0roo56hfnufk5
- 原文链接：https://x.com/kimmonismus/status/2085822646705696784

## AI 摘要

OpenAI 称其即将推出的 Astra 模型网络安全能力可能已达“严重”阈值，初步内部评估发现其在智能体编码和网络性能上取得重大进展，因此无法排除达到该级别的可能。OpenAI 将暂停 Astra 中未满足强化安全控制要求的内部活动（含部分训练），同时限制网络和工具访问、加强模型权重保护，并监控所有智能体用途的风险行为。整体开发将在更严格管控下继续，这可能为中国模型带来契机。

## 正文

Parts of the internal work on Astra - possibly including certain training activities - will be paused until they meet the new security requirements. Overall development will continue under stricter controls.

Could be a leverage and an opportunity for chinese models.

### 引用推文

> Chubby♨️：Doesnt look good for a soon to come releas for GPT-Astra: OpenAI says its upcoming Astra model may have reached the "Critical" threshold for cybersecurity capab...