feat(ms-ai-architect): R7.4 bølge 2 — payload-11..19 dømt+ingestet (153 claims, 9 flagged, 41 flagg; ledger 153→162) [skip-docs]
This commit is contained in:
parent
8e4c8b59c7
commit
3043ff74d4
1 changed files with 592 additions and 1 deletions
|
|
@ -4,7 +4,7 @@
|
|||
"derived_from": "scripts/kb-update/data/full-pass-worklist.json (243 due, all never-verified)",
|
||||
"cadence": "R7–R10 (5 økter × ~49, 8–10 samtidige)",
|
||||
"batch": "R7.1",
|
||||
"count": 153,
|
||||
"count": 162,
|
||||
"generated": "2026-07-18"
|
||||
},
|
||||
"files": [
|
||||
|
|
@ -7651,6 +7651,597 @@
|
|||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/digital-samhandling-eif-5-layers.md",
|
||||
"batch": "R7.4",
|
||||
"judged_at": "2026-07-31",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 19,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-governance/norwegian-public-sector-governance/digital-samhandling-eif-5-layers.md#1",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/foundry/concepts/foundry-models-overview",
|
||||
"evidence_quote": "Some of the details available in the model card are: 1. Quick facts: Key information about the model at a quick glance 2. Details tab: Detailed information about the model, like description, version info, and supported data type 3. Deployments tab: A list of existing deployments for the model 4. Benchmarks tab: Performance benchmark metrics for select models 5. License tab: Legal information related to model licensing",
|
||||
"reason": "Model catalog finnes, men den kanoniske siden som ville enumerert kapabilitetene beskriver den som hub for oppdagelse, filtrering, benchmarking og distribusjon — verken modellregister eller lineage tracking er nevnt noe sted; søk på Foundry-lineage gir kun modellversjonering, så fraværet er bevis mot påstanden.",
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/digital-samhandling-eif-5-layers.md",
|
||||
"line": 109,
|
||||
"claim": "Microsoft Foundry har en model catalog som brukes til modellregister og lineage tracking.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/norwegian-public-sector-governance/digital-samhandling-eif-5-layers.md#6",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R3",
|
||||
"evidence_url": "https://learn.microsoft.com/power-platform/guidance/coe/governance-components",
|
||||
"evidence_quote": "The Power Platform CoE Starter Kit is no longer actively maintained. Its core capabilities are part of the Power Platform admin center.",
|
||||
"reason": "Rammen claimen forutsetter er erstattet: kitet vedlikeholdes ikke lenger og kapabilitetene er flyttet til Power Platform admin center, og de enumererte governance-komponentene er compliance-flyter for apper, flyter, chatboter og koblinger — ikke AI governance-arbeidsflyter (dokumentasjonen sier til og med at Copilot Studio-agenter i Teams-miljøer ikke er synlige i kitet).",
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/digital-samhandling-eif-5-layers.md",
|
||||
"line": 134,
|
||||
"claim": "Power Platform CoE Starter Kit inneholder AI governance workflows.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/norwegian-public-sector-governance/digital-samhandling-eif-5-layers.md#10",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/foundry/openai/concepts/retired-models",
|
||||
"evidence_quote": "The following models are retired and no longer available for use or for new deployments. ... o1-preview | — | — | July 28, 2025 | o1",
|
||||
"reason": "GPT-4o-halvdelen holder (den står i Current models), men o1-preview ble pensjonert 28. juli 2025 og er ikke lenger tilgjengelig — én bærende del er motsagt, og da faller hele claimen.",
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/digital-samhandling-eif-5-layers.md",
|
||||
"line": 143,
|
||||
"claim": "Azure OpenAI Service tilbyr API for modellene GPT-4o | o1-preview.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/norwegian-public-sector-governance/digital-samhandling-eif-5-layers.md#14",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/ai-services/content-safety/overview",
|
||||
"evidence_quote": "Content Safety provides several APIs for different moderation needs: Prompt Shields ... Groundedness detection (preview) ... Protected material text detection ... Task adherence API",
|
||||
"reason": "Moderasjonshalvdelen holder, men red teaming er ikke en Content Safety-kapabilitet: den kanoniske enumerasjonen lister kun moderasjons-/beskyttelses-API-er, mens red teaming ligger i Foundrys AI Red Teaming Agent, som selv omtaler Azure AI Content Safety filters som et separat produksjonsrekkverk.",
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/digital-samhandling-eif-5-layers.md",
|
||||
"line": 149,
|
||||
"claim": "Azure AI Content Safety dekker moderation | red teaming.",
|
||||
"disposition": "outdated"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/forvaltningsloven-ai-decisions.md",
|
||||
"batch": "R7.4",
|
||||
"judged_at": "2026-07-31",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 16,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-governance/norwegian-public-sector-governance/forvaltningsloven-ai-decisions.md#1",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/machine-learning/concept-responsible-ai-dashboard",
|
||||
"evidence_quote": "The Responsible AI dashboard brings together, in a comprehensive view, various new and pre-existing tools. The dashboard integrates these tools with Azure Machine Learning CLI v2, Azure Machine Learning Python SDK v2, and Azure Machine Learning studio.",
|
||||
"reason": "Transparency Notes og model cards er dokumentert for Azure AI-tjenester, men den kanoniske siden plasserer Responsible AI Dashboard i Azure Machine Learning (CLI/SDK/studio), ikke i Azure AI Services — én bærende del av produktmappingen er dermed feil.",
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/forvaltningsloven-ai-decisions.md",
|
||||
"line": 106,
|
||||
"claim": "Azure AI Services tilbyr verktøyene Responsible AI Dashboard | Model Cards | Transparency Notes, som kan brukes som utgangspunkt for dokumentasjonskravet.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/norwegian-public-sector-governance/forvaltningsloven-ai-decisions.md#3",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/foundry/responsible-ai/openai/data-privacy",
|
||||
"evidence_quote": "The abuse monitoring data store where prompts and completions are stored for human review is logically separated by customer resource (each request includes the resource ID of the customer's Foundry resource).",
|
||||
"reason": "Claimet påstår zero data retention som standardegenskap (ingen lagret data), men den autoritative data-privacy-siden sier at prompts og completions lagres for menneskelig gjennomgang, at Responses API/Threads/Stored completions persisterer innhold, og at bortfall av lagring krever godkjent modified abuse monitoring.",
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/forvaltningsloven-ai-decisions.md",
|
||||
"line": 137,
|
||||
"claim": "Azure OpenAI har zero data retention (ingen lagret data), noe som sikrer personvern men gjør sporing og forklarbarhet vanskeligere.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/norwegian-public-sector-governance/forvaltningsloven-ai-decisions.md#8",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/api-management/api-management-howto-app-insights",
|
||||
"evidence_quote": "Application Insights is designed to assess application performances using statistical analysis. It's not: Intended to be an audit system. Suited for logging each individual request for high-volume APIs.",
|
||||
"reason": "Claimet påstår full request/response-logging, men Learn sier at Application Insights ikke er ment som revisjonssystem eller egnet til å logge hver enkelt forespørsel, at standardverdien for Number of payload bytes to log er 0, og at maksimal telemetristørrelse er 64 KB.",
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/forvaltningsloven-ai-decisions.md",
|
||||
"line": 249,
|
||||
"claim": "Azure Application Insights gir full request/response-logging.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/norwegian-public-sector-governance/forvaltningsloven-ai-decisions.md#9",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/foundry-classic/concepts/prompt-flow",
|
||||
"evidence_quote": "Prompt flow in Microsoft Foundry and Azure Machine Learning will be retired on April 20, 2027. Prompt flow is no longer recommended for new development.",
|
||||
"reason": "RAI-verktøy og modellstyring i Foundry er dokumentert, men den bærende delen prompt flow for menneske-i-sløyfen holder ikke: Learn beskriver prompt flow som et utviklingsverktøy for prototyping og iterasjon, ikke som en menneske-i-sløyfen-mekanisme, og det er avviklingsmerket og frarådet for ny utvikling.",
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/forvaltningsloven-ai-decisions.md",
|
||||
"line": 267,
|
||||
"claim": "Microsoft Foundry tilbyr komplett RAI-verktøysett | model governance | prompt flow for menneske-i-sløyfen.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/norwegian-public-sector-governance/forvaltningsloven-ai-decisions.md#12",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/microsoft-copilot-studio/guidance/autonomous-agents",
|
||||
"evidence_quote": "Instead of responding only in conversations, they operate continuously in the background—monitoring data, reacting to conditions, and running workflows at scale.",
|
||||
"reason": "Menneske-i-sløyfen er dokumentert, men den bærende avgrensningen kun dialog-/samtalebaserte løsninger motsies direkte: autonome agenter kjører hendelsesdrevet i bakgrunnen, og computer use utvider Copilot Studio utover samtalelogikk.",
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/forvaltningsloven-ai-decisions.md",
|
||||
"line": 271,
|
||||
"claim": "Copilot Studio har menneske-i-sløyfen innebygd og dekker kun dialog-/samtalebaserte løsninger.",
|
||||
"disposition": "outdated"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/gevinstrealisering-ai-projects.md",
|
||||
"batch": "R7.4",
|
||||
"judged_at": "2026-07-31",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 8,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-governance/norwegian-public-sector-governance/gevinstrealisering-ai-projects.md#1",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/power-platform/guidance/adoption/business-value",
|
||||
"evidence_quote": "",
|
||||
"reason": "Påstanden gjelder DFØ og Prosjektveiviserens gevinstkategorier; den siterte siden nevner verken DFØ eller Prosjektveiviseren, og oppslag mot learn.microsoft.com finner ingen side som omtaler dem — siden lister fire value drivers (Performance improvement, Direct or indirect cost savings, Risk mitigation, Business transformation) for Power Platform-løsninger, ikke DFØs taksonomi, så kilden kan verken bekrefte eller motbevise påstanden.",
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/gevinstrealisering-ai-projects.md",
|
||||
"line": 49,
|
||||
"claim": "DFØ og Prosjektveiviseren opererer med tre hovedkategorier av gevinster: Ytelsesforbedrende gevinster (Performance improvement) | Kostnadsbesparelser (Direct/indirect cost savings) | Risikoreduksjon (Risk mitigation).",
|
||||
"disposition": "unsourced"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/norge-ai-strategy-government.md",
|
||||
"batch": "R7.4",
|
||||
"judged_at": "2026-07-31",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 15,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-governance/norwegian-public-sector-governance/norge-ai-strategy-government.md#3",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R3",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/azure-sovereign-clouds/microsoft-sovereign-cloud",
|
||||
"evidence_quote": "Microsoft Sovereign Cloud, formerly Microsoft Cloud for Sovereignty, is an evolution and expansion of the original offering.",
|
||||
"reason": "Claimets organiserende ramme er erstattet — tilbudet heter nå Microsoft Sovereign Cloud — og ingen Learn-side sier at data under dette tilbudet lagres i Norge Øst og Norge Vest; suverenitetssidene rammer inn residens som EU Data Boundary og kundevalgte regioner.",
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/norge-ai-strategy-government.md",
|
||||
"line": 126,
|
||||
"claim": "Med Microsoft Cloud for Sovereignty lagres data i norske Azure-regioner: Norge Øst og Norge Vest.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/norwegian-public-sector-governance/norge-ai-strategy-government.md#15",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R3",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/foundry/what-is-foundry",
|
||||
"evidence_quote": "| Brand | Azure AI Studio / Azure AI Foundry | Microsoft Foundry |",
|
||||
"reason": "Claimet navngir Azure AI Studio som verktøyet, men Foundry-siden fører opp Azure AI Studio som forrige merkenavn og Microsoft Foundry som gjeldende — rammen er erstattet, selv om Copilot Studio Analytics-delen står ved lag (reaksjoner og kommentarer vises i agentens Analytics-fane).",
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/norge-ai-strategy-government.md",
|
||||
"line": 223,
|
||||
"claim": "Brukerfeedback-løkker for AI-løsninger settes opp med Azure AI Studio og Copilot Studio Analytics.",
|
||||
"disposition": "outdated"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/nsm-grunnprinsipper-ai-mapping.md",
|
||||
"batch": "R7.4",
|
||||
"judged_at": "2026-07-31",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 13,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-governance/norwegian-public-sector-governance/nsm-grunnprinsipper-ai-mapping.md#1",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/ai-foundry/concepts/rbac-azure-ai-foundry",
|
||||
"evidence_quote": "The Foundry RBAC roles were recently renamed. Foundry User, Foundry Owner, Foundry Account Owner, and Foundry Project Manager were previously named Azure AI User, Azure AI Owner, Azure AI Account Owner, and Azure AI Project Manager.",
|
||||
"reason": "Den kanoniske RBAC-siden lister de innebygde rollene Foundry Agent Consumer, Foundry User, Foundry Project Manager, Foundry Account Owner og Foundry Owner; ingen rolle heter Model Deployer, og Azure AI Developer frarådes eksplisitt for Foundry-arbeid.",
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/nsm-grunnprinsipper-ai-mapping.md",
|
||||
"line": 131,
|
||||
"claim": "Azure RBAC tilbyr granulære AI-roller: AI Developer | AI User | Model Deployer.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/norwegian-public-sector-governance/nsm-grunnprinsipper-ai-mapping.md#2",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/microsoft-copilot-studio/guidance/sec-gov-phase3",
|
||||
"evidence_quote": "Assign the Environment Maker role to users who need to author agents.",
|
||||
"reason": "Den kanoniske tabellen over predefinerte sikkerhetsroller lister System Administrator, System Customizer, Environment Maker, Bot Transcript Viewer, Bot Author (deprecated), Bot Contributor og Omnichannel Administrator; trioen Publisher/Author/Viewer finnes ikke i oppregningen.",
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/nsm-grunnprinsipper-ai-mapping.md",
|
||||
"line": 133,
|
||||
"claim": "Copilot Studio har sikkerhetsrollene Publisher | Author | Viewer.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/norwegian-public-sector-governance/nsm-grunnprinsipper-ai-mapping.md#9",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/search-reliability",
|
||||
"evidence_quote": "Because AI Search isn't a primary data storage solution, it doesn't provide self-service backup and restore options.",
|
||||
"reason": "Den kanoniske reliability-siden sier at tjenesten ikke tilbyr backup/restore og at AI Search er en enkeltregion-tjeneste; ingen geo-redundant backup av vektordata er oppgitt noe sted.",
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/nsm-grunnprinsipper-ai-mapping.md",
|
||||
"line": 264,
|
||||
"claim": "Azure AI Search har geo-redundant backup av vektordata.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/norwegian-public-sector-governance/nsm-grunnprinsipper-ai-mapping.md#10",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure-region-availability",
|
||||
"evidence_quote": "| **Model** | **Version** | **francecentral** | **germanywestcentral** | **norwayeast** | **polandcentral** | **spaincentral** | **swedencentral** | **switzerlandnorth** | **uksouth** | **westeurope** | ... | gpt-4o | 2024-11-20 | ✅ | - | ✅ | - | - | ✅ | ✅ | ✅ | - |",
|
||||
"reason": "Standard/Regional-tabellen for Europa fører norwayeast som egen distribusjonsregion for Azure OpenAI (gpt-4o 2024-11-20, text-embedding-3-large, ada-002, whisper), så den bærende delen «Norge (via Sweden)» er feil - Norge har egen regional tilgjengelighet og trenger ikke gå via Sverige.",
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/nsm-grunnprinsipper-ai-mapping.md",
|
||||
"line": 515,
|
||||
"claim": "Azure OpenAI kan geo-pinnes til Sverige, Norge (via Sweden) eller andre EU-regioner.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/norwegian-public-sector-governance/nsm-grunnprinsipper-ai-mapping.md#11",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/security/benchmark/azure/introduction",
|
||||
"evidence_quote": "The Microsoft cloud security benchmark v2 (preview) is now available.",
|
||||
"reason": "Den siterte siden oppgir v1 (med v1-baselines) og v2 (preview) som nyeste versjon; det finnes ingen MCSB v3.0, så den andre bærende halvdelen av påstanden er motsagt selv om v1-delen holder.",
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/nsm-grunnprinsipper-ai-mapping.md",
|
||||
"line": 545,
|
||||
"claim": "Microsoft cloud security benchmark (MCSB) v1.0 er gjeldende standard, og v3.0 er også tilgjengelig for nyere baselines.",
|
||||
"disposition": "outdated"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/ros-analyse-ai-systems.md",
|
||||
"batch": "R7.4",
|
||||
"judged_at": "2026-07-31",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 17,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-governance/norwegian-public-sector-governance/ros-analyse-ai-systems.md#3",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/defender-for-cloud/ai-threat-protection",
|
||||
"evidence_quote": "Release state | Generally available (GA)",
|
||||
"reason": "Deteksjonsdelen holder (alerts for jailbreak, credential theft, access anomaly), men den løftebærende statusdelen er motsagt: AI threat protection er GA, ikke preview — og planen heter Defender for AI Services, ikke Defender for AI Workloads.",
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/ros-analyse-ai-systems.md",
|
||||
"line": 313,
|
||||
"claim": "Defender for AI Workloads i Microsoft Defender for Cloud er i preview, og detekterer unormale AI-interaksjoner.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/norwegian-public-sector-governance/ros-analyse-ai-systems.md#4",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/security/fundamentals/threat-detection",
|
||||
"evidence_quote": "Microsoft Defender for AI Services - Provides runtime protection for AI workloads against jailbreaks, data exposure, and suspicious access patterns.",
|
||||
"reason": "JIT-access og threat intelligence er bekreftet, men den kanoniske opplistingen av Defender for Cloud-planer navngir Microsoft Defender for AI Services; ingen hentet side omtaler en plan ved navn «Defender for AI Workloads» i brødteksten (kun «AI workloads» som beskrivende ord) — det navngitte elementet mangler i listen som ville enumerert det.",
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/ros-analyse-ai-systems.md",
|
||||
"line": 313,
|
||||
"claim": "Microsoft Defender for Clouds AI-relevante kapabiliteter omfatter Defender for AI Workloads | Just-in-Time (JIT) access | Threat intelligence.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/norwegian-public-sector-governance/ros-analyse-ai-systems.md#9",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/security/develop/threat-modeling-tool-feature-overview",
|
||||
"evidence_quote": "Reports | Create HTML reports to share with others.",
|
||||
"reason": "To av fire løftebærende deler holder (trust boundaries via Trust line/Border boundary-stensiler, automatisk trusselgenerering fra dataflytdiagrammet), men den fullstendige meny-/funksjonsopplistingen har ingen import av Azure-arkitektur (modeller tegnes manuelt fra Azure-stensiler) og ingen eksport til Azure DevOps — kun HTML-rapporter og deling via OneDrive.",
|
||||
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/ros-analyse-ai-systems.md",
|
||||
"line": 343,
|
||||
"claim": "Microsoft Threat Modeling Tool støtter import av Azure-arkitektur (Microsoft Foundry, Copilot Studio) | identifisering av trust boundaries | automatisk generering av trusler basert på dataflyt | eksport til Azure DevOps for sporing av tiltak.",
|
||||
"disposition": "outdated"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-center-of-excellence-setup.md",
|
||||
"batch": "R7.4",
|
||||
"judged_at": "2026-07-31",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 20,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-governance/responsible-ai/ai-center-of-excellence-setup.md#1",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/agents/center-of-excellence/",
|
||||
"evidence_quote": "Choose your CoE structure and operating model: Choose a centralized, hybrid, or federated model based on your patterns and maturity.",
|
||||
"reason": "Den kanoniske AI-CoE-siden enumererer TRE driftsmodeller (Centralized, Hybrid, Federated) — verken Unified eller Decentralized finnes der, og den siterte CAF-CoE-siden kontrasterer bare sentralisert vs. rådgivende; den ordrette fire-veis-taksonomien (centralized, unified, federated, or decentralized) finnes kun på Fabric adoption roadmap for data-/analyse-COE, ikke for AI CoE.",
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-center-of-excellence-setup.md",
|
||||
"line": 52,
|
||||
"claim": "Microsoft anbefaler fire strukturmodeller for AI Center of Excellence: Sentralisert CoE | Unified CoE | Federated CoE | Desentralisert CoE.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/responsible-ai/ai-center-of-excellence-setup.md#2",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/cloud-adoption-framework/scenarios/ai/center-of-excellence",
|
||||
"evidence_quote": "Include senior data scientists, machine learning engineers, AI governance experts, AI security specialists, and AI operations professionals in the team.",
|
||||
"reason": "Den kanoniske siden navngir CoE-leder, business leaders og fem tekniske roller, men Data Engineer er fraværende i enumereringen — claimet påstår åtte roller inkludert en rolle kilden ikke lister.",
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-center-of-excellence-setup.md",
|
||||
"line": 112,
|
||||
"claim": "AI CoE-teamet består av åtte roller: AI CoE Leader | Senior Data Scientist | ML Engineer | AI Governance Expert | AI Security Specialist | AI Operations Professional | Business Leader | Data Engineer.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/responsible-ai/ai-center-of-excellence-setup.md#5",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/cloud-adoption-framework/scenarios/ai/center-of-excellence",
|
||||
"evidence_quote": "Watch for approval delays and knowledge bottlenecks where AI experts in the CoE can't support all teams.",
|
||||
"reason": "Kilden angir TRE signaler (approval delays, knowledge bottlenecks, prioriteringsfriksjon med produktteam) — et fjerde signal Policy compliance finnes ikke i enumereringen, verken her eller på den agentiske CoE-sidens Signals it is time to shift.",
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-center-of-excellence-setup.md",
|
||||
"line": 263,
|
||||
"claim": "Microsoft angir fire signaler (inflection points) for overgang fra sentralisert CoE til advisory-modell: Approval delays | Knowledge bottlenecks | Priority friction | Policy compliance.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/responsible-ai/ai-center-of-excellence-setup.md#6",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/cloud-adoption-framework/scenarios/ai/govern",
|
||||
"evidence_quote": "",
|
||||
"reason": "Den siterte govern-siden inneholder ingen modenhetsskala, og Maturity Model for Microsoft 365 ligger ikke lenger på learn.microsoft.com (URL-en 301-redirigerer til pnp.github.io), så nivånavnene Managed/Predictable/Optimizing kan verken bekreftes eller motbevises mot Learn; andre Learn-modeller (Fabric, Power Platform, agentisk AI) er egne modeller og motbeviser ikke denne.",
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-center-of-excellence-setup.md",
|
||||
"line": 274,
|
||||
"claim": "Modenhetsnivåene (Microsoft 365 Maturity Model tilpasset Azure AI) har fem nivåer: 100 - Initial | 200 - Managed | 300 - Defined | 400 - Predictable | 500 - Optimizing.",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/responsible-ai/ai-center-of-excellence-setup.md#9",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R3",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/cloud-adoption-framework/scenarios/ai/platform/governance",
|
||||
"evidence_quote": "Use Azure Policy to apply built-in policy definitions for: Microsoft Foundry, Foundry Tools, Azure AI Search",
|
||||
"reason": "Claimets fire-tjeneste-ramme er erstattet: CAF enumererer nå TRE mål, der Azure AI Services opptrer som Foundry Tools (lenken peker på /azure/ai-services/policy-reference) og Azure OpenAI er foldet inn i Microsoft Foundry Models — det finnes ikke lenger en frittstående Azure OpenAI-oppføring ved siden av Microsoft Foundry.",
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-center-of-excellence-setup.md",
|
||||
"line": 506,
|
||||
"claim": "Cloud Adoption Framework peker på nøkkelpolicyer for AI innenfor fire tjenester: Microsoft Foundry (model deployment restrictions, content filter enforcement) | Azure AI Services (allowed SKUs, network isolation) | Azure AI Search (encryption, network security) | Azure OpenAI (model restrictions, content filtering).",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/responsible-ai/ai-center-of-excellence-setup.md#12",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/cloud-adoption-framework/scenarios/ai/govern",
|
||||
"evidence_quote": "",
|
||||
"reason": "Verken den siterte govern-siden, CAF-CoE-siden eller CAF Plan (Access AI resources) kobler lisenstyper til CoE-roller; ingen learn.microsoft.com-side enumererer lisensbehov per CoE-rolle, så verdiene kan verken bekreftes eller motbevises.",
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-center-of-excellence-setup.md",
|
||||
"line": 570,
|
||||
"claim": "Lisensbehov per CoE-rolle: CoE Lead = M365 E5 + Azure subscription contributor | Data Scientist = M365 E3 + Microsoft Foundry + VS Enterprise | ML Engineer = M365 E3 + Azure DevOps + GitHub Copilot | AI Governance Expert = M365 E5 Compliance + Purview | Security Specialist = M365 E5 Security + Defender for Cloud.",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/responsible-ai/ai-center-of-excellence-setup.md#13",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/cloud-adoption-framework/scenarios/ai/",
|
||||
"evidence_quote": "Diagram that shows the 6 phases of AI adoption: Strategy, Plan, Ready, Govern, Secure, Manage.",
|
||||
"reason": "Andre halvdel holder (AI CoE er steg 1 under Manage AI operations: Establish an AI center of excellence for strategic guidance), men rekkefølgen er en feil load-bearing del: fire uavhengig hentede CAF-sider (strategy, plan, govern, manage) og adopsjonssjekklisten setter Secure FØR Manage, mens claimet påstår Manage → Secure.",
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-center-of-excellence-setup.md",
|
||||
"line": 658,
|
||||
"claim": "Cloud Adoption Framework sitt AI-scenario har oppdatert struktur i seks steg: Strategy → Plan → Ready → Govern → Manage → Secure, og AI CoE er referert under «Manage AI operations» steg 1.",
|
||||
"disposition": "outdated"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-ethics-in-public-sector.md",
|
||||
"batch": "R7.4",
|
||||
"judged_at": "2026-07-31",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 20,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-governance/responsible-ai/ai-ethics-in-public-sector.md#2",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/cloud-adoption-framework/scenarios/ai/govern",
|
||||
"evidence_quote": "This article helps you establish an organizational process for governing AI. You use this guidance to integrate AI risk management into your broader risk management strategies, creating a unified approach to AI, cybersecurity, and privacy governance.",
|
||||
"reason": "Den kanoniske CAF-siden for AI-styring beskriver fire prosessteg (risikovurdering, policy, håndheving, overvåking) og ingen governance-roller; verken AI Governance Board, DPO eller AI Ethics Committee forekommer, og CAFs eneste navngitte organ er AI Center of Excellence.",
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-ethics-in-public-sector.md",
|
||||
"line": 79,
|
||||
"claim": "Microsoft Cloud Adoption Framework anbefaler fire governance-roller for AI: AI Governance Board | AI Center of Excellence | Data Protection Officer (DPO) | AI Ethics Committee.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/responsible-ai/ai-ethics-in-public-sector.md#4",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/machine-learning/how-to-responsible-ai-scorecard",
|
||||
"evidence_quote": "An Azure Machine Learning Responsible AI scorecard is a PDF report that's generated based on Responsible AI dashboard insights and customizations to accompany your machine learning models.",
|
||||
"reason": "Den kanoniske scorecard-siden beskriver utelukkende PDF (PDF-rapport, nedlasting av PDF, View all PDF scorecards); HTML mangler helt i oppregningen, så påstanden om PDF eller HTML er motsagt av fraværet.",
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-ethics-in-public-sector.md",
|
||||
"line": 101,
|
||||
"claim": "Responsible AI Scorecard kan genereres som PDF eller HTML for innsynsforespørsler.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/responsible-ai/ai-ethics-in-public-sector.md#8",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R7",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/machine-learning/concept-fairness-ml",
|
||||
"evidence_quote": "Disparity in model performance: These sets of metrics calculate the disparity (difference) in the values of the selected performance metric across subgroups of data.",
|
||||
"reason": "Siden navngir to klasser dispariteter (disparity in model performance og disparity in selection rate) og paritetsbetingelsene demographic parity, equalized odds, equal opportunity og bounded group loss; verken disparate impact ratio eller equal opportunity difference er metrikker siden navngir, og metrikknavnene er selve påstanden.",
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-ethics-in-public-sector.md",
|
||||
"line": 144,
|
||||
"claim": "Fairness Assessment i Responsible AI Dashboard måler disparate impact ratio og equal opportunity difference.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/responsible-ai/ai-ethics-in-public-sector.md#11",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/foundry-classic/ai-services/content-safety-overview",
|
||||
"evidence_quote": "Protected material detection for text: Identifies protected text material, such as known song lyrics, articles, or other content.",
|
||||
"reason": "Den kanoniske Foundry-siden for Content Safety ramser opp minst ni kapabiliteter (tekstmoderering, groundedness detection, protected material for tekst og for kode, prompt shields, bilde- og multimodal moderering, egendefinerte kategorier, safety system message), så det oppgitte antallet tre er motsagt, og Content Safety Studio er en portalflate snarere enn en av kapabilitetene.",
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-ethics-in-public-sector.md",
|
||||
"line": 204,
|
||||
"claim": "Microsoft Foundry tilbyr tre Responsible AI-kapabiliteter: Content Safety Studio | Prompt Shields | Groundedness Detection.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/responsible-ai/ai-ethics-in-public-sector.md#20",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/power-platform/admin/powerapps-flow-licensing-faq",
|
||||
"evidence_quote": "Licensed by tenant, Microsoft Copilot Studio entitles a tenant to 25,000 messages per month.",
|
||||
"reason": "Power Platform-lisensdokumentasjonen kjenner Power Apps Premium og Power Automate Premium, men ingen SKU som heter Power Platform Premium, og Copilot Studio lisensieres separat per tenant - altså ikke inkludert i en Premium-plan.",
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-ethics-in-public-sector.md",
|
||||
"line": 339,
|
||||
"claim": "Power Platform Premium inkluderer AI Builder | Copilot Studio | Premium connectors.",
|
||||
"disposition": "outdated"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-risk-taxonomy-classification.md",
|
||||
"batch": "R7.4",
|
||||
"judged_at": "2026-07-31",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 25,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-governance/responsible-ai/ai-risk-taxonomy-classification.md#1",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/legal/ai-code-of-conduct",
|
||||
"evidence_quote": "",
|
||||
"reason": "Den siterte siden sier kun at den er utformet for å samsvare bedre med nye AI-reguleringer (EU AI Act), men oppgir ingen firetrinns risikotaksonomi, og søk fant ingen learn.microsoft.com-side som lister unacceptable/high/limited/minimal.",
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-risk-taxonomy-classification.md",
|
||||
"line": 31,
|
||||
"claim": "Microsofts risikotaksonomi er tilpasset EU AI Acts fire hovedkategorier: unacceptable | high | limited | minimal.",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/responsible-ai/ai-risk-taxonomy-classification.md#3",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/legal/ai-code-of-conduct",
|
||||
"evidence_quote": "",
|
||||
"reason": "Siden krever menneskelig tilsyn ved beslutninger med konsekvenser for rettslig/økonomisk stilling og arbeidsmuligheter, men enumererer ingen høyrisiko-kategorier som critical infrastructure, credit scoring, healthcare diagnosis eller biometric identification, og nevner ikke impact assessment; ingen learn.microsoft.com-side oppgir denne listen.",
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-risk-taxonomy-classification.md",
|
||||
"line": 47,
|
||||
"claim": "Kategorien High Risk krever strict compliance, human oversight og impact assessment, og omfatter: critical infrastructure | employment decisions | credit scoring | healthcare diagnosis | biometric identification.",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/responsible-ai/ai-risk-taxonomy-classification.md#13",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-artificial-intelligence-security",
|
||||
"evidence_quote": "Define critical actions: Identify high-risk AI operations requiring human review such as external data transfers, processing of confidential information, or decisions impacting financial or operational outcomes, using risk assessments to prioritize review pathways.",
|
||||
"reason": "AI-5 er den kanoniske seksjonen som ville listet HITL-scenariene, og den gir tre eksempeltyper uten terskler og uten skillet mandatory approval/mandatory review; healthcare diagnosis/treatment, employment decisions og legal/compliance decisions finnes ikke i AI-5 i det hele tatt, så den påståtte femdelte kravlisten er ikke i kilden.",
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-risk-taxonomy-classification.md",
|
||||
"line": 205,
|
||||
"claim": "MCSB-kontrollen AI-5 (human-in-the-loop) krever HITL for fem scenarier: external data transfer (mandatory approval) | financial transactions over terskel (mandatory approval) | healthcare diagnosis/treatment (mandatory review) | employment decisions (mandatory review) | legal/compliance decisions (mandatory approval).",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/responsible-ai/ai-risk-taxonomy-classification.md#15",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R1",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/ai-foundry/concepts/ai-red-teaming-agent",
|
||||
"evidence_quote": "Hateful and Unfair Content ... Sexual Content ... Violent Content ... Self-Harm-Related Content ... Protected Materials ... Code vulnerability ... Ungrounded attributes ... Prohibited actions ... Sensitive data leakage ... Task adherence",
|
||||
"reason": "Supported risk categories-tabellen lister ti risikokategorier, ikke fire; taket claimet setter (fire kategorier) er passert, selv om de fire navngitte inngår blant de ti.",
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-risk-taxonomy-classification.md",
|
||||
"line": 221,
|
||||
"claim": "AI Red Teaming Agent utfører automatisert adversarial testing mot fire risikokategorier: Violence | Hate | Sexual | Self-Harm.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/responsible-ai/ai-risk-taxonomy-classification.md#19",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R1",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/machine-learning/concept-responsible-ai-dashboard",
|
||||
"evidence_quote": "Data analysis ... Model overview and fairness assessment ... Error analysis ... Model interpretability ... Counterfactual what-if ... Causal analysis",
|
||||
"reason": "Responsible AI dashboard components-seksjonen lister seks komponenter, ikke fire; claimet utelater Data analysis og Counterfactual what-if, så antallet det oppgir er passert av kilden.",
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-risk-taxonomy-classification.md",
|
||||
"line": 268,
|
||||
"claim": "Responsible AI Dashboard i Azure Machine Learning består av fire komponenter: Error Analysis | Fairness Assessment | Model Explainability | Causal Inference.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/responsible-ai/ai-risk-taxonomy-classification.md#20",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure-region-availability",
|
||||
"evidence_quote": "| **Model** | **Version** | **francecentral** | **germanywestcentral** | **italynorth** | **norwayeast** | **polandcentral** | **spaincentral** | **swedencentral** | **switzerlandnorth** | **switzerlandwest** | **uksouth** | **westeurope** |",
|
||||
"reason": "Jeg sjekket den kanoniske region-tilgjengelighetssiden for Foundry-modeller: norwayeast står i alle Europa-tabellene (ni forekomster), mens norwaywest ikke forekommer én eneste gang - Norway East-delen holder, men Norway West som region for AI-arbeidslast gjør det ikke, og én bærende del som svikter felles hele claimet.",
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-risk-taxonomy-classification.md",
|
||||
"line": 304,
|
||||
"claim": "Norsk data residency for AI-arbeidslast oppnås i Azure-regionene Norway East og Norway West.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/responsible-ai/ai-risk-taxonomy-classification.md#21",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/ai-foundry/concepts/ai-red-teaming-agent",
|
||||
"evidence_quote": "",
|
||||
"reason": "AI Red Teaming Agent-siden dokumenterer kapabilitet, støttede risikokategorier og regioner, men sier ingenting om lisensiering eller at funksjonen er inkludert uten tillegg; søk fant ingen learn.microsoft.com-side som oppgir dette, og pris-/lisensvilkår ligger på azure.microsoft.com-prissidene.",
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-risk-taxonomy-classification.md",
|
||||
"line": 316,
|
||||
"claim": "Red teaming (AI Red Teaming Agent) er inkludert i Microsoft Foundry / AI Foundry uten separat lisens eller tillegg.",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/responsible-ai/ai-risk-taxonomy-classification.md#23",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/security/security-for-ai/security-dashboard-for-ai",
|
||||
"evidence_quote": "Supported products: Microsoft Entra ... Microsoft Defender ... Microsoft Purview ... Microsoft Security Copilot",
|
||||
"reason": "Supported products-tabellen er den kanoniske enumerasjonen av hva dashboardet avhenger av, og den navngir de fire produktene uten én eneste lisensvariant - hverken Entra ID P2, Defender for Cloud P2, Purview Compliance eller per capacity unit for Security Copilot forekommer på siden, og artikkelen sier tvert imot at dashboardet ikke krever egen lisens ut over de underliggende sikkerhetsproduktene.",
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-risk-taxonomy-classification.md",
|
||||
"line": 324,
|
||||
"claim": "Security Dashboard for AI er avhengig av fire lisensierte produkter: Microsoft Entra ID P2 (identity risk detection) | Microsoft Defender for Cloud P2 (AI workload threat protection) | Microsoft Purview Compliance (AI-accessible data classification) | Security Copilot (lisensiert per capacity unit, AI risk exploration via prompts).",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-governance/responsible-ai/ai-risk-taxonomy-classification.md#25",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R1",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-artificial-intelligence-security",
|
||||
"evidence_quote": "AI-6: Establish monitoring and detection ... AI-7: Perform continuous AI Red Teaming",
|
||||
"reason": "Artikkelen bærer kontrollseksjonene AI-1 til og med AI-7, ikke AI-1 til AI-5; three-pillar-modellen stemmer, men den oppgitte rekkevidden er passert av kilden.",
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/ai-risk-taxonomy-classification.md",
|
||||
"line": 434,
|
||||
"claim": "MCSB v2-artikkelen Artificial Intelligence Security inneholder sikkerhetskontrollene AI-1 til AI-5 samt three-pillar-modellen.",
|
||||
"disposition": "outdated"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-governance/references/responsible-ai/content-safety-implementation.md",
|
||||
"batch": "R7.1",
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue