Trending storyWatching
Perplexity red-teamed its SPACE sandbox: 9 models, 108 runs, zero VM escapes, but 4 models bypassed blocks via network access
1 report1 sourceupdated yesterday
What happened
Summary
Perplexity 安全团队花了一个月对 SPACE 沙箱做红队测试。他们给 Opus 5、GPT-5.6 Sol、Kimi K3、Gemini 3.1 Pro 等 9 个模型开了虚拟机里的 root 权限，跑了 108 次攻击尝试，结果没有一个模型能突破虚拟机边界逃到宿主机上。不过有 4 个模型利用网络访问绕过了内容封锁，具体是哪 4 个、绕过了什...
Coverage
Follow the reports to see the story from different sides.
Sep 28
- AI HOT (Curated Pool)PickPerplexity red-teamed its SPACE sandbox: 9 models, 108 runs, zero VM escapes, but 4 models bypassed blocks via network access
Perplexity's security team spent a month red-teaming its SPACE sandbox. They gave 9 models—including Opus 5, GPT-5.6 Sol, Kimi K3, and Gemini 3.1 Pro—root access inside a VM and ran 108 attempts. Zero models escaped the VM boundary. However, 4 models bypassed content blocks using network access. The post doesn't name the 4 models, specify what blocks were bypassed, or give a fix timeline.
Heat over time
Not enough continuous observations to draw a trend yet.