# Perplexity finds red-teaming setup can steer agents away from tested security boundary

> Original title: We are publishing new research on our findings from red-teaming SPACE, our VM sandbox for Computer agents.

- Source: Perplexity
- Published: 2026-11-11T18:21:22.000Z
- Bad Influence: https://badinfluence.news/items/amiezs4qveffbzrka5mbobqun
- Original: https://x.com/perplexity_ai/status/2120582674549203004

## Summary

Perplexity publishes research from red-teaming SPACE, its VM sandbox for Computer agents, testing 12 models on VM escape and network policy bypass tasks. It finds the standard Capture the Flag setup used in red-teaming can steer agents away from the security boundary being tested. It also finds that security-specialized models do not guarantee better results.

## Post

We are publishing new research on our findings from red-teaming SPACE, our VM sandbox for Computer agents.

We gave 12 models two tasks: escape the VM to take a credential from the host, and bypass SPACE's network policy to reach a blocked URL.

We found that the standard Capture the Flag setup used in red-teaming can steer agents away from the security boundary being tested, and that security-specialized models don't guarantee better results.

https://www.perplexity.ai/hub/blog/escaping-space-part-ii

![](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Fmedia%2FHW3StEBaIAAKK1W.png&mode=thumb&exp=1794787200&sig=66fdbac940578856)