--- title: How safe are gpt-oss-safeguard models? | daily.dev description: Cisco researchers evaluated the security posture of OpenAI's gpt-oss and gpt-oss-safeguard models (20b and 120b variants) against adversarial prompt injection... canonical: https://daily.dev/posts/how-safe-are-gpt-oss-safeguard-models--an6wpmrgq twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: How safe are gpt-oss-safeguard models? | daily.dev og:description: Cisco researchers evaluated the security posture of OpenAI's gpt-oss and gpt-oss-safeguard models (20b and 120b variants) against adversarial prompt injection... og:url: https://daily.dev/posts/how-safe-are-gpt-oss-safeguard-models--an6wpmrgq og:image: https://api.daily.dev/og/posts/aN6WpmRgq.png og:image:alt: How safe are gpt-oss-safeguard models? 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. # How safe are gpt-oss-safeguard models? **[Cisco](https://daily.dev/sources/cisco)** · 7 min read · 1 upvotes · 0 comments ## Summary Cisco researchers evaluated the security posture of OpenAI's gpt-oss and gpt-oss-safeguard models (20b and 120b variants) against adversarial prompt injection and jailbreak attacks. Key findings: multi-turn attacks are the dominant threat, causing attack success rates to jump 5x–8.5x compared to single-turn attacks across all variants. The 120b standard model showed the strongest single-turn resilience, while safeguard variants sometimes introduced exploitable complexity. Model size proved a stronger predictor of baseline resilience than safeguard tuning. Since these are open-weight models, anyone can deploy them in chat settings regardless of OpenAI's intended use case. The analysis recommends a defense-in-depth approach including real-time monitoring, content filtering, rate limiting, and continuous red-teaming rather than relying on model selection alone. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Similar posts on daily.dev - [When AI safety constrains defenders more than attackers](https://daily.dev/posts/when-ai-safety-constrains-defenders-more-than-attackers-n5jrhk0br) · CSO Online · 0 upvotes · 0 comments --- Tags: [#openai](https://daily.dev/tags/openai), [#ai-safety](https://daily.dev/tags/ai-safety), [#prompt-injection](https://daily.dev/tags/prompt-injection) [View this post on daily.dev](https://daily.dev/posts/how-safe-are-gpt-oss-safeguard-models--an6wpmrgq) ```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":"How safe are gpt-oss-safeguard models?","url":"https://daily.dev/posts/how-safe-are-gpt-oss-safeguard-models--an6wpmrgq","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/how-safe-are-gpt-oss-safeguard-models--an6wpmrgq"},"datePublished":"2026-02-18T23:05:39.097Z","dateModified":"2026-02-18T23:06:03.176Z","description":"Cisco researchers evaluated the security posture of OpenAI's gpt-oss and gpt-oss-safeguard models (20b and 120b variants) against adversarial prompt injection...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/ba6732b2b8c190d7acb6ae542ef50c3b?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/ba6732b2b8c190d7acb6ae542ef50c3b?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Cisco","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":"Cisco","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/68b8a8be17184235a5067d7c84162f77","url":"https://daily.dev/sources/cisco"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/how-safe-are-gpt-oss-safeguard-models--an6wpmrgq","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":1},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"openai,ai-safety,prompt-injection","timeRequired":"PT7M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Cisco","item":"https://daily.dev/sources/cisco"},{"@type":"ListItem","position":3,"name":"How safe are gpt-oss-safeguard models?"}]} ```