--- title: Why your AI agent needs two sandboxes: Benchmark data description: Benchmark results from testing two sandboxing approaches for AI coding agents on Red Hat OpenShift: Kata micro-VMs (OpenShift sandboxed containers) and NVIDIA... canonical: https://daily.dev/posts/why-your-ai-agent-needs-two-sandboxes-benchmark-data-zexdlcp4n twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Why your AI agent needs two sandboxes: Benchmark data | daily.dev og:description: Benchmark results from testing two sandboxing approaches for AI coding agents on Red Hat OpenShift: Kata micro-VMs (OpenShift sandboxed containers) and NVIDIA... og:url: https://daily.dev/posts/why-your-ai-agent-needs-two-sandboxes-benchmark-data-zexdlcp4n og:image: https://api.daily.dev/og/posts/ZeXdlCp4n.png og:image:alt: Why your AI agent needs two sandboxes: Benchmark data og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Why your AI agent needs two sandboxes: Benchmark data **[Red Hat Developer](https://daily.dev/sources/rhdev)** · 12 min read · 0 upvotes · 0 comments ## Summary Benchmark results from testing two sandboxing approaches for AI coding agents on Red Hat OpenShift: Kata micro-VMs (OpenShift sandboxed containers) and NVIDIA OpenShell. Three pod configurations were tested against two attacks — a prompt injection data exfiltration attempt and a real kernel exploit (CVE-2026-31431, a Linux page-cache write primitive). Kata-only stopped the kernel exploit but leaked data via exfiltration. OpenShell-only blocked exfiltration but was bypassed by the kernel exploit. Only the dual configuration (both layers) stopped both attacks. The post argues AI agents uniquely expose both threat surfaces simultaneously — untrusted input (prompt injection) and arbitrary command execution (kernel attack surface) — making layered defense-in-depth necessary rather than optional. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Similar posts on daily.dev - [Layered sandboxing for AI agents: OpenShift and OpenShell](https://daily.dev/posts/layered-sandboxing-for-ai-agents-openshift-and-openshell-lekdk6l8t) · Red Hat Developer · 0 upvotes · 0 comments - [AI sandboxing is having its Kubernetes moment](https://daily.dev/posts/ai-sandboxing-is-having-its-kubernetes-moment-v0gh4b1qc) · CNCF · 37 upvotes · 1 comments - [Every layer counts: Defense in depth for AI agents with Red Hat AI](https://daily.dev/posts/every-layer-counts-defense-in-depth-for-ai-agents-with-red-hat-ai-afirrw4f5) · Red Hat Developer · 1 upvotes · 0 comments - [The Untrusted Autonomous Workload and AI Sandboxes](https://daily.dev/posts/the-untrusted-autonomous-workload-and-ai-sandboxes-cotb1iovu) · Docker · 4 upvotes · 0 comments --- Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#containers](https://daily.dev/tags/containers), [#openshift](https://daily.dev/tags/openshift), [#prompt-injection](https://daily.dev/tags/prompt-injection) [View this post on daily.dev](https://daily.dev/posts/why-your-ai-agent-needs-two-sandboxes-benchmark-data-zexdlcp4n) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Why your AI agent needs two sandboxes: Benchmark data","url":"https://daily.dev/posts/why-your-ai-agent-needs-two-sandboxes-benchmark-data-zexdlcp4n","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/why-your-ai-agent-needs-two-sandboxes-benchmark-data-zexdlcp4n"},"datePublished":"2026-07-23T07:16:10.837Z","dateModified":"2026-07-23T07:16:31.598Z","description":"Benchmark results from testing two sandboxing approaches for AI coding agents on Red Hat OpenShift: Kata micro-VMs (OpenShift sandboxed containers) and NVIDIA...","image":"https://media.daily.dev/image/upload/s--2-1xRawN--/f_auto/v1722860399/public/Placeholder%2011","thumbnailUrl":"https://media.daily.dev/image/upload/s--2-1xRawN--/f_auto/v1722860399/public/Placeholder%2011","isAccessibleForFree":true,"articleSection":"Red Hat Developer","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Red Hat Developer","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/2a8895f3216c4900bdc455585a4ad3e4","url":"https://daily.dev/sources/rhdev"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/why-your-ai-agent-needs-two-sandboxes-benchmark-data-zexdlcp4n","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-agents,containers,openshift,prompt-injection","timeRequired":"PT12M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Red Hat Developer","item":"https://daily.dev/sources/rhdev"},{"@type":"ListItem","position":3,"name":"Why your AI agent needs two sandboxes: Benchmark data"}]} ```