open/ms-ai-architect
1
0
Fork 0
Code Issues Pull requests Releases Activity

feat(ms-ai-architect): R7.4 bølge 1 — payload-01..10 dømt+ingestet (165 claims, 9 flagged/1 pass, 21 flagg; ledger 143→153) [skip-docs]

Browse source
This commit is contained in:
Kjell Tore Guttormsen 2026-07-31 15:49:17 +02:00
commit a7cb4b2bb3
1 changed files with 362 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

363
scripts/kb-eval/data/judge-pass-manifest.json
View file

@ -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": 143,
"count": 153,
"generated": "2026-07-18"
},
"files": [
@ -7210,6 +7210,53 @@
}
]
},
{
"file": "skills/ms-ai-governance/references/monitoring-observability/response-quality-metrics-rag.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/monitoring-observability/response-quality-metrics-rag.md#1",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/azure/foundry/concepts/evaluation-evaluators/rag-evaluators",
"evidence_quote": "Groundedness Pro (preview) | System evaluation | You want a strict groundedness definition powered by Azure AI Content Safety and use our service model",
"reason": "Den kanoniske RAG-evaluator-siden merker to av seks RAG-evaluatorer som (preview) — Groundedness Pro og Response Completeness — så et blankt GA-stempel for RAG-evaluering er motsagt; til sammenligning sier textual-similarity-siden eksplisitt All six evaluators are generally available (GA) der det faktisk stemmer.",
"file": "skills/ms-ai-governance/references/monitoring-observability/response-quality-metrics-rag.md",
"line": 4,
"claim": "Response quality metrics og evaluering for RAG-systemer i Microsoft Foundry / Azure AI Evaluation har status GA.",
"disposition": "outdated"
},
{
"id": "ms-ai-governance/monitoring-observability/response-quality-metrics-rag.md#2",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/azure/foundry/concepts/built-in-evaluators",
"evidence_quote": "General purpose evaluators | Evaluator | Purpose | Coherence | Measures logical consistency and flow of responses. | Fluency | Measures natural language quality and readability.",
"reason": "Coherence og Fluency er kategorisert som general purpose-evaluatorer, ikke RAG-evaluatorer; RAG-listen på siden er Retrieval, Document Retrieval, Groundedness, Groundedness Pro, Relevance og Response Completeness, så claimens to siste ledd er feil kategorisert.",
"file": "skills/ms-ai-governance/references/monitoring-observability/response-quality-metrics-rag.md",
"line": 37,
"claim": "De AI-assisterte RAG-spesifikke evaluatorene i Azure AI Evaluation er: Groundedness | Groundedness Pro | Relevance | Response Completeness | Retrieval | Coherence | Fluency.",
"disposition": "outdated"
},
{
"id": "ms-ai-governance/monitoring-observability/response-quality-metrics-rag.md#18",
"judge_verdict": "not_grounded",
"rule": "R2",
"evidence_url": "https://learn.microsoft.com/azure/foundry/concepts/observability",
"evidence_quote": "Evaluations in the agents playground are enabled by default for all Foundry projects and are included in consumption-based billing.",
"reason": "Den kanoniske pricing-seksjonen for observability oppgir ingen gratis tier; den sier tvert imot at evalueringsfunksjoner (playground-evalueringer og risiko-/sikkerhetsevalueringer) faktureres etter forbruk via egen Foundry Observability-prisside, altså utover de underliggende GPT-kallene.",
"file": "skills/ms-ai-governance/references/monitoring-observability/response-quality-metrics-rag.md",
"line": 524,
"claim": "Microsoft Foundry har en gratis tier for evaluerings-UI-et; man betaler kun for underliggende compute (GPT-kall).",
"disposition": "outdated"
}
]
},
{
"file": "skills/ms-ai-governance/references/monitoring-observability/security-and-audit-logging-ai.md",
"batch": "R7.1",
@ -7233,6 +7280,172 @@
}
]
},
{
"file": "skills/ms-ai-governance/references/monitoring-observability/sla-monitoring-ai-services.md",
"batch": "R7.4",
"judged_at": "2026-07-31",
"per_file_verdict": "flagged",
"claim_count": 23,
"verified": null,
"verified_by": null,
"flags": [
{
"id": "ms-ai-governance/monitoring-observability/sla-monitoring-ai-services.md#2",
"judge_verdict": "source_silent",
"rule": "",
"evidence_url": "https://learn.microsoft.com/en-us/azure/foundry-classic/openai/faq#what-are-the-slas-service-level-agreements-in-azure-openai",
"evidence_quote": "",
"reason": "FAQ-en bekrefter at Availability SLA gjelder alle ressurser, men ingen learn.microsoft.com-side oppgir prosentsatsen 99.9% - den bærende tallverdien finnes kun på den eksterne SLA-siden på azure.microsoft.com, og motsies ikke.",
"file": "skills/ms-ai-governance/references/monitoring-observability/sla-monitoring-ai-services.md",
"line": 37,
"claim": "Availability SLA for Azure OpenAI garanterer 99.9% oppetid og gjelder alle Azure OpenAI-ressurser.",
"disposition": "unsourced"
},
{
"id": "ms-ai-governance/monitoring-observability/sla-monitoring-ai-services.md#7",
"judge_verdict": "not_grounded",
"rule": "R2",
"evidence_url": "https://learn.microsoft.com/en-us/azure/azure-monitor/alerts/alerts-types",
"evidence_quote": "Unlike traditional log search alerts that aggregate rows over a defined period, simple log alerts evaluate each row individually.",
"reason": "Per-rad-evaluering og nær-sanntid er bekreftet, men ingen av de kanoniske sidene (alerts-overview, alerts-types, alerts-create-simple-alert) merker simple log search alerts som preview, mens de samme sidene eksplisitt merker andre funksjoner som public preview - statuspåstanden er dermed motsagt av fraværet.",
"file": "skills/ms-ai-governance/references/monitoring-observability/sla-monitoring-ai-services.md",
"line": 50,
"claim": "Simple Log Search Alerts er i preview og evaluerer hver logg-rad individuelt (nær-sanntid), raskere enn tradisjonelle log alerts.",
"disposition": "outdated"
},
{
"id": "ms-ai-governance/monitoring-observability/sla-monitoring-ai-services.md#8",
"judge_verdict": "not_grounded",
"rule": "R7",
"evidence_url": "https://learn.microsoft.com/en-us/azure/foundry/openai/monitor-openai-reference",
"evidence_quote": "For Azure OpenAI latency monitoring, use Time to Response (AzureOpenAITimeToResponse), Time to Last Byte (AzureOpenAITTLTInMS), Time Between Tokens (AzureOpenAINormalizedTBTInMS), or Normalized Time to First Byte (AzureOpenAINormalizedTTFTInMS).",
"reason": "Ingen metrikk heter TimeToFirstToken eller ContextTokens; siden navngir AzureOpenAINormalizedTTFTInMS/TimeToResponse for første-token-latens og ProcessedPromptTokens/InputTokens for input-tokens, så to av de oppgitte metrikknavnene er feil strenger.",
"file": "skills/ms-ai-governance/references/monitoring-observability/sla-monitoring-ai-services.md",
"line": 56,
"claim": "Azure Monitor-metrikkene for SLA-tracking av Azure OpenAI er AvailabilityRate | ModelAvailabilityRate | ModelRequests | StatusCode (dimensjon) | TimeToFirstToken | ContextTokens | GeneratedTokens.",
"disposition": "outdated"
},
{
"id": "ms-ai-governance/monitoring-observability/sla-monitoring-ai-services.md#12",
"judge_verdict": "not_grounded",
"rule": "R7",
"evidence_url": "https://learn.microsoft.com/en-us/azure/azure-monitor/reference/supported-metrics/microsoft-cognitiveservices-accounts-metrics",
"evidence_quote": "Normalized Time to First Byte Time it takes for first byte of response data to be received after request is made by model, normalized by token. Applies to PTU, PTU-managed, and Pay-as-you-go deployments.",
"reason": "Det finnes ingen metrikk som heter TimeToFirstToken (nærmeste er NormalizedTimeToFirstToken/AzureOpenAINormalizedTTFTInMS og TimeToResponse), og latensmetrikkene aggregeres som Maximum, Minimum, Average - ikke P95/P99.",
"file": "skills/ms-ai-governance/references/monitoring-observability/sla-monitoring-ai-services.md",
"line": 62,
"claim": "TimeToFirstToken måler latens til første token ved streaming, knyttes til Latency SLA for PTU-deployments, og aggregeres som P95/P99.",
"disposition": "outdated"
},
{
"id": "ms-ai-governance/monitoring-observability/sla-monitoring-ai-services.md#16",
"judge_verdict": "source_silent",
"rule": "",
"evidence_url": "https://learn.microsoft.com/en-us/azure/ai-foundry/openai/how-to/business-continuity-disaster-recovery",
"evidence_quote": "",
"reason": "Ingen learn-side oppgir 99.9% for Azure OpenAI eller peker på West Europe som failover-region for Norway East; BCDR-siden omtaler kun failover mellom regioner generelt og lenker til den eksterne SLA-siden.",
"file": "skills/ms-ai-governance/references/monitoring-observability/sla-monitoring-ai-services.md",
"line": 271,
"claim": "Azure OpenAIs 99.9% SLA gjelder for Norway East-regionen, og West Europe kan brukes som failover-region ved lengre utfall.",
"disposition": "unsourced"
},
{
"id": "ms-ai-governance/monitoring-observability/sla-monitoring-ai-services.md#18",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/en-us/azure/azure-monitor/alerts/alerts-types",
"evidence_quote": "Query-based metric alerts are now in public preview for alerting based on Prometheus and OpenTelemetry metrics.",
"reason": "De seks alarmtypene stemmer med tabellen, men ingen kanonisk side merker simple log search alerts som preview - samme sider merker derimot eksplisitt query-based metric alerts som public preview - så den oppgitte preview-statusen er en angitt, bærende del som ikke holder.",
"file": "skills/ms-ai-governance/references/monitoring-observability/sla-monitoring-ai-services.md",
"line": 389,
"claim": "Azure Monitor har alert-typene metric | log search | simple log search (preview, per-rad evaluering) | activity log | smart detection | Prometheus.",
"disposition": "outdated"
},
{
"id": "ms-ai-governance/monitoring-observability/sla-monitoring-ai-services.md#20",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/en-us/azure/azure-monitor/alerts/alerts-overview",
"evidence_quote": "Alert frequency of equal to or more than 5 minutes: While the condition continues to be met, a notification is sent between the configured frequency and double the frequency. For example, for an alert rule with a frequency of 15 minutes, a notification is sent sometime between 15 to 30 minutes.",
"reason": "1-6 minutter for frekvens under 5 minutter stemmer, men regelen for 5 minutter eller mer er frekvens til dobbel frekvens - 15-30 minutter er kun eksempelet for 15 minutters frekvens, ikke den generelle verdien (en 5-minutters regel gir 5-10 minutter).",
"file": "skills/ms-ai-governance/references/monitoring-observability/sla-monitoring-ai-services.md",
"line": 389,
"claim": "Stateless metric alerts med evalueringsfrekvens under 5 minutter trigges 1-6 minutter etter, mens frekvens på 5 minutter eller mer trigges 15-30 minutter etter.",
"disposition": "outdated"
},
{
"id": "ms-ai-governance/monitoring-observability/sla-monitoring-ai-services.md#21",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/en-us/azure/azure-monitor/alerts/alerts-overview",
"evidence_quote": "1 minute: The alert condition isn't met for 10 minutes. 5 to 15 minutes: The alert condition isn't met for three frequency periods. 15 minutes to 11 hours: The alert condition isn't met for two frequency periods. 11 to 12 hours: The alert condition isn't met for one frequency period.",
"reason": "Tabellen har fire rader, og to av dem faller utenfor det påståtte intervallet 2-3 frekvensperioder: 1 minutts frekvens krever 10 minutter, og 11-12 timers frekvens krever kun én periode.",
"file": "skills/ms-ai-governance/references/monitoring-observability/sla-monitoring-ai-services.md",
"line": 389,
"claim": "Stateful log search alerts settes til resolved når betingelsen ikke er oppfylt i 2-3 frekvensperioder (avhengig av frekvens).",
"disposition": "outdated"
}
]
},
{
"file": "skills/ms-ai-governance/references/monitoring-observability/token-usage-tracking-attribution.md",
"batch": "R7.4",
"judged_at": "2026-07-31",
"per_file_verdict": "flagged",
"claim_count": 22,
"verified": null,
"verified_by": null,
"flags": [
{
"id": "ms-ai-governance/monitoring-observability/token-usage-tracking-attribution.md#5",
"judge_verdict": "not_grounded",
"rule": "R2",
"evidence_url": "https://learn.microsoft.com/cli/azure/cognitiveservices/account",
"evidence_quote": "az cognitiveservices account update ... --tags Space-separated tags: key[=value] [key[=value] ...].",
"reason": "Den kanoniske CLI-referansen har ingen kommandogruppe az openai; tagging av Azure OpenAI-ressurser gjøres med az cognitiveservices account update --name --resource-group --tags, så den påståtte kommandoen finnes ikke.",
"file": "skills/ms-ai-governance/references/monitoring-observability/token-usage-tracking-attribution.md",
"line": 182,
"claim": "Azure OpenAI-ressurser kan tagges for kostnadsattribuering med Azure CLI-kommandoen az openai account update --name <navn> --resource-group <rg> --tags.",
"disposition": "outdated"
},
{
"id": "ms-ai-governance/monitoring-observability/token-usage-tracking-attribution.md#6",
"judge_verdict": "not_grounded",
"rule": "R2",
"evidence_url": "https://learn.microsoft.com/azure/foundry/openai/how-to/quota",
"evidence_quote": "Quota is assigned to your subscription on a per-region, per-model, per-deployment-type basis in units of Tokens-per-Minute (TPM).",
"reason": "Den kanoniske kvotesiden beskriver én samlet TPM-grense (pluss RPM) der estimatet inkluderer prompt og max_tokens; den lister ingen separate ITPM/OTPM-grenser for Azure OpenAI (ITPM/OTPM finnes kun for andre tilbud som Databricks Foundation Model APIs og Claude-modeller i Foundry).",
"file": "skills/ms-ai-governance/references/monitoring-observability/token-usage-tracking-attribution.md",
"line": 241,
"claim": "Azure OpenAI har separate token-baserte rate limits for input tokens per minutt (ITPM) og output tokens per minutt (OTPM).",
"disposition": "outdated"
},
{
"id": "ms-ai-governance/monitoring-observability/token-usage-tracking-attribution.md#12",
"judge_verdict": "not_grounded",
"rule": "R7",
"evidence_url": "https://learn.microsoft.com/en-us/azure/foundry-classic/openai/concepts/use-your-data",
"evidence_quote": "This is the `topNDocuments` parameter in the API, and is 5 by default.",
"reason": "Siden bekrefter knappene og standardverdiene (5 og 1024) samt inScope og strictness, men API-parameteren heter topNDocuments, ikke retrieved_document_count; parameternavnet er selve påstanden (R7s load-bearing-unntak), og navnet finnes ikke på Learn.",
"file": "skills/ms-ai-governance/references/monitoring-observability/token-usage-tracking-attribution.md",
"line": 400,
"claim": "Token-bruk i Azure OpenAI On Your Data justeres via parameterne retrieved_document_count (default 5) | chunk_size (default 1024) | strictness | inScope.",
"disposition": "outdated"
},
{
"id": "ms-ai-governance/monitoring-observability/token-usage-tracking-attribution.md#15",
"judge_verdict": "not_grounded",
"rule": "R2",
"evidence_url": "https://learn.microsoft.com/azure/azure-monitor/reference/supported-metrics/microsoft-cognitiveservices-accounts-metrics",
"evidence_quote": "Processed FineTuned Training HoursNumber of Training Hours Processed on an OpenAI FineTuned Model | `FineTunedTrainingHours`",
"reason": "Den kanoniske enumereringen av metrikker for Microsoft.CognitiveServices/accounts inneholder FineTunedTrainingHours, men verken FineTuningHours eller FineTuningTokens finnes i listen.",
"file": "skills/ms-ai-governance/references/monitoring-observability/token-usage-tracking-attribution.md",
"line": 445,
"claim": "Azure Monitor eksponerer metrikkene FineTuningHours og FineTuningTokens for Azure OpenAI under ResourceProvider MICROSOFT.COGNITIVESERVICES.",
"disposition": "outdated"
}
]
},
{
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/accessibility-requirements-wcag-norway.md",
"batch": "R7.1",
@ -7290,6 +7503,154 @@
"verified_by": "judge-v3.1",
"flags": []
},
{
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/citizen-communication-ai-decisions.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/citizen-communication-ai-decisions.md#15",
"judge_verdict": "not_grounded",
"rule": "R2",
"evidence_url": "https://learn.microsoft.com/en-us/azure/ai-foundry/openai/concepts/abuse-monitoring",
"evidence_quote": "Some customers may want to use Models sold by Azure for a use case that involves the processing of highly sensitive or highly confidential data, or otherwise may conclude that they don't want or don't have the right to permit Microsoft to store and conduct human review on their prompts and completions for abuse detection. To address these concerns, Microsoft allows customers who meet additional Limited Access eligibility criteria to apply to modify abuse monitoring by completing this form.",
"reason": "Jeg hentet de kanoniske sidene som enumererer hvordan lagring av prompts/completions kan slås av (abuse-monitoring og data-privacy) — ingen av dem navngir noen «Zero Data Retention»-modus for Azure OpenAI; den dokumenterte mekanismen heter modified abuse monitoring under Limited Access, og data-privacy-sidens «no prompts or completions are stored» gjelder modellen (stateless), ikke tjenesten, som per standard lagrer for abuse monitoring.",
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/citizen-communication-ai-decisions.md",
"line": 139,
"claim": "Azure OpenAI med Zero Data Retention lagrer ikke prompts og completions, og tjenesten gjenbruker dem ikke",
"disposition": "outdated"
}
]
},
{
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/copyright-ai-training-data-norway.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/copyright-ai-training-data-norway.md#15",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure?pivots=azure-openai#fine-tuning-models",
"evidence_quote": "Global training provides more affordable training per token, but doesn't offer data residency. It's currently available to Foundry resources in the following regions: ... Norway East ...",
"reason": "Norway East står KUN i listen over Global training-regioner, som eksplisitt ikke gir data residency; fine-tuning-tabellens kolonne «Standard regions» (den eneste treningstypen med data residency) lister i Europa bare Sweden Central (og Switzerland West for gpt-35-turbo), ikke Norway East — den bærende delen «inkludert Norway East» er motsagt av kilden.",
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/copyright-ai-training-data-norway.md",
"line": 190,
"claim": "Microsoft tilbyr data residency i flere europeiske regioner, inkludert Norway East, for fine-tuning-operasjoner.",
"disposition": "outdated"
}
]
},
{
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/digdir-ai-governance-structure.md",
"batch": "R7.4",
"judged_at": "2026-07-31",
"per_file_verdict": "pass",
"claim_count": 12,
"verified": "2026-07-31",
"verified_by": "judge-v3.1",
"flags": []
},
{
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/digdir-principle-1-user-centric-design.md",
"batch": "R7.4",
"judged_at": "2026-07-31",
"per_file_verdict": "flagged",
"claim_count": 7,
"verified": null,
"verified_by": null,
"flags": [
{
"id": "ms-ai-governance/norwegian-public-sector-governance/digdir-principle-1-user-centric-design.md#4",
"judge_verdict": "source_silent",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure",
"evidence_quote": "",
"reason": "Modellkatalogen fører fortsatt gpt-4 (turbo-2024-04-09) som generelt tilgjengelig, men ingen Learn-side oppgir noe MMLU-resultat for GPT-4 — Foundrys ledertavler bruker MMLU-Pro, ikke MMLU — så den løftende MMLU-delen kan verken bekreftes eller motbevises.",
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/digdir-principle-1-user-centric-design.md",
"line": 237,
"claim": "GPT-4 er en tilgjengelig modell som kan score høyt på MMLU-benchmarken.",
"disposition": "unsourced"
}
]
},
{
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/digdir-principle-2-interoperability.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/digdir-principle-2-interoperability.md#8",
"judge_verdict": "not_grounded",
"rule": "R1",
"evidence_url": "https://learn.microsoft.com/en-us/azure/logic-apps/logic-apps-overview",
"evidence_quote": "The Azure Logic Apps platform provides 1,400+ prebuilt connectors that you can use to integrate your workflows with various services, systems, apps, and data.",
"reason": "Den siterte arkitektursiden oppgir ingen konnektortelling, så jeg fulgte til den kanoniske Logic Apps-siden: den sier 1,400+, mens claimet oppgir en nedre grense på over 400 - 3,5x over R1s 2x-terskel og beslutningsendrende, siden konnektorbredde nettopp er utvelgelseskriteriet i en interoperabilitets-referanse, så bindingen villeder materielt.",
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/digdir-principle-2-interoperability.md",
"line": 113,
"claim": "Azure Logic Apps har over 400 konnektorer for både skytjenester og on-premises-systemer.",
"disposition": "outdated"
}
]
},
{
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/digdir-principle-4-trust-security.md",
"batch": "R7.4",
"judged_at": "2026-07-31",
"per_file_verdict": "flagged",
"claim_count": 22,
"verified": null,
"verified_by": null,
"flags": [
{
"id": "ms-ai-governance/norwegian-public-sector-governance/digdir-principle-4-trust-security.md#16",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/en-us/azure/ai-services/cognitive-services-virtual-networks",
"evidence_quote": "You can limit access to your resources with request filtering, which allows requests that originate only from specified IP addresses, IP ranges, or from a list of subnets in Azure Virtual Networks.",
"reason": "Managed Identity og Private Endpoints holder, men den kanoniske nettverkssiden som eksplisitt dekker Azure OpenAI regner opp service endpoints, IP-regler og private endepunkter - VNET-injection er fraværende; den dokumenterte network injection gjelder Foundry Agent Service-compute i et subnett delegert til Microsoft.App/environments, ikke Azure OpenAI-inferensendepunktet, så én bærende del av påstanden feiler.",
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/digdir-principle-4-trust-security.md",
"line": 213,
"claim": "Azure OpenAI støtter Managed Identity-autentisering | VNET-injection | Private Endpoints.",
"disposition": "outdated"
}
]
},
{
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/digital-accessibility-action-plan.md",
"batch": "R7.4",
"judged_at": "2026-07-31",
"per_file_verdict": "flagged",
"claim_count": 14,
"verified": null,
"verified_by": null,
"flags": [
{
"id": "ms-ai-governance/norwegian-public-sector-governance/digital-accessibility-action-plan.md#8",
"judge_verdict": "not_grounded",
"rule": "R2",
"evidence_url": "https://learn.microsoft.com/azure/ai-services/translator/language-support",
"evidence_quote": "Norwegian Bokmål | `nb` | ✔ | ✔ | ✔ | ✔ | ✔ | ✔ |",
"reason": "Jeg leste hele den kanoniske Translation-tabellen (Afrikaans til Zulu, ~145 språk) på Translators language-support-side: den oppfører Norwegian Bokmål (`nb`), men INGEN oppføring for Norwegian Nynorsk finnes noe sted på siden — fraværet i den enumererende kanoniske kilden er bevis mot eksistenspåstanden (R2). «100+ språk» og «bokmål» holder, men nynorsk-delen er bærende (norsk offentlig sektor har språkkrav) og er feil, så hele claimet faller (R8).",
"file": "skills/ms-ai-governance/references/norwegian-public-sector-governance/digital-accessibility-action-plan.md",
"line": 310,
"claim": "Azure AI Translator støtter 100+ språk, inkludert nynorsk og bokmål.",
"disposition": "outdated"
}
]
},
{
"file": "skills/ms-ai-governance/references/responsible-ai/content-safety-implementation.md",
"batch": "R7.1",