# Microsoft Purview vs IBM Guardium: which do AI models recommend for DSPM, October 2026

IT AI Recommendation Index, October 2026 Edition, Data security posture management. Zero of fourteen models named Microsoft Purview first on the direct prompt; zero named IBM Guardium. Page: https://it-ai-index.com/security/data-security-posture-management/microsoft-purview-vs-ibm-guardium/

| | First-choice share | Rank | Negative rate | Labels | Models naming it |
|---|---|---|---|---|---|
| Microsoft Purview | 18% | #1 of 12 | 11% | 37 | 14 of 14 |
| IBM Guardium | 4% | #6 of 12 | 36% | 11 | 8 of 14 |

## The direct prompt, model by model

- GPT-5.4 mini: neither first, one named (first choices: Varonis Data Security Platform) (alternatives: Microsoft Purview, Teleskope)
- GLM 4.7 FlashX: neither first, one named (first choices: Securiti, Varonis Data Security Platform) (alternatives: Cyera, Microsoft Purview)
- MiniMax M2.5: neither first, one named (first choices: Cyera, Sentra) (alternatives: Concentric, Forcepoint DSPM, Microsoft Purview)
- GPT-6 Luna: neither first, one named (first choices: Wiz DSPM) (alternatives: Cyera, Microsoft Purview, Sentra)
- Claude Haiku 4.5: neither named (first choices: Cyera) (alternatives: Netwrix, Normalyze, Sentra)
- Gemini 3.5 Flash: neither named (first choices: Wiz) (alternatives: CrowdStrike, Cyera, Proofpoint DSPM, Sentra)
- Perplexity Sonar: neither named (first choices: Sentra) (alternatives: Concentric, Cyera, Prisma Cloud DSPM, Wiz)
- Grok 4.1 Fast: neither named (first choices: Sentra) (alternatives: Concentric AI, Cyera, Varonis Data Security Platform)
- Mistral Small: neither named (first choices: Sentra) (alternatives: Concentric, Cyera)
- DeepSeek V4 Flash: neither named (first choices: Sentra) (alternatives: Concentric AI, Cyera)
- Llama 4 Maverick: neither named (first choices: Concentric, Cyera, Sentra)
- Qwen 3.7 Flash: neither named (first choices: Wiz) (alternatives: Cyera, Microsoft Defender for Cloud, Orca Security)
- Kimi K2: neither named (first choices: Wiz) (alternatives: BigID, Varonis Data Security Platform)
- Muse Glimmer 30B: neither named (first choices: Sentra) (alternatives: Concentric AI, Cyera)

## What the models said about Microsoft Purview

- "is outstanding for Microsoft-heavy environments ... but is notoriously limited when dealing with AWS, GCP, or non-Microsoft SaaS applications" (Gemini 3.5 Flash, negative prompt, soft negative)
- "organizations that don't already have M365 E5 licensing will find that advanced DSPM capabilities come at significant additional cost" (Claude Haiku 4.5, budget prompt, soft negative)
- "Start with Microsoft Purview Suite ($10-12/user/month) if you're already in the Microsoft ecosystem—it offers the best balance of features, support, and predictable costs." (Kimi K2, budget prompt, first choice)

## What the models said about IBM Guardium

- "miss the discovery, posture management, access governance, and behavioral detection that define true DSPM platforms" (Claude Haiku 4.5, negative prompt, hard negative)
- "cannot afford massive, complex enterprise tools (like BigID, IBM Guardium, or Symantec DLP)" (Gemini 3.5 Flash, paraphrase prompt, hard negative)
- "far more accessible than enterprise-heavy tools like IBM Guardium (custom pricing, G2 4.3/5 but enterprise-focused)" (Grok 4.1 Fast, paraphrase prompt, soft negative)
- "IBM Guardium Data Protection and Safetica are recommended sensitive data discovery and protection tools for a mid-sized B2B company." (Llama 4 Maverick, paraphrase prompt, first choice)
- "Primary choice: IBM Guardium Data Protection – the most complete, highest‑rated solution" (GLM 4.7 FlashX, paraphrase prompt, first choice)
- "Choose IBM Guardium DSPM if you want a cloud-native, agentless approach across hybrid cloud and SaaS." (Perplexity Sonar, comparative prompt, alternative)

Share is the count of first choices across the direct, paraphrase, budget and scale prompts over all fourteen models, for a mid-market B2B company; rank is within the category. Comparisons are drawn for the top eight products in each category. Published under CC BY 4.0; the output is the models' output, and nothing here is a recommendation by the index.