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

feat(ms-ai-architect): R7.5 bølge 2 — payload-11..20 dømt+ingestet (209 claims, 51 flagg; ledger 202→212) [skip-docs]

Browse source 
Ti dommere (R75J-11..20) kjørt parallelt med live MS Learn-fetch per claim.
Alle ti returnerte, validerte 10/10 mot slot + claim-id-sett, og ingen
foldet ASCII — verifisert mot BEGGE dialektene (ae/oe og enkelt-vokal a/o),
0 treff. Mot 2/10 foldet i bølge 1 med samme modell og effort; forskjellen
er den skjerpede prompten med eksempler på begge foldemønstre.

Ingen pass-record i bølgen, så ingen re-judge-gate og ingen stempling —
manifestet er eneste endrede fil.

- 11 model-fingerprinting-watermarking.md ......... 21 claims,  7 flagg
- 12 norwegian-content-safety.md .................. 23 claims,  5 flagg
- 13 output-validation-grounding-verification.md .. 19 claims,  3 flagg
- 14 owasp-llm-top10-azure-mitigations.md ......... 18 claims,  2 flagg
- 15 pii-detection-norwegian-context.md ........... 19 claims,  4 flagg
- 16 prompt-injection-defense-patterns.md ......... 14 claims,  3 flagg
- 17 security-copilot-integration.md .............. 34 claims,  9 flagg
- 18 security-scoring-rubrics-6x5.md .............. 21 claims,  4 flagg
- 19 supply-chain-security-ai-models.md ........... 17 claims,  9 flagg
- 20 ai-builder-credits-transition.md ............. 23 claims,  5 flagg

Ingest dry-run kryssjekket mot egen telling fil for fil før ekte ingest.
Judge-dato 2026-08-01 (reell dato, ÅS#1) lagt inn for slot 11-20.

Ledger 202→212 records (199 flagged + 13 pass), 803→854 flagg.
R7.5: 20/51 dømt. Suite 947/947.
This commit is contained in:
Kjell Tore Guttormsen 2026-08-01 20:21:06 +02:00
commit 8ac64e407a
1 changed files with 723 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

724
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": 202,
"count": 212,
"generated": "2026-07-18"
},
"files": [
@ -11546,6 +11546,657 @@
}
]
},
{
"file": "skills/ms-ai-security/references/ai-security-engineering/model-fingerprinting-watermarking.md",
"batch": "R7.5",
"judged_at": "2026-08-01",
"per_file_verdict": "flagged",
"claim_count": 21,
"verified": null,
"verified_by": null,
"flags": [
{
"id": "ms-ai-security/ai-security-engineering/model-fingerprinting-watermarking.md#3",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/en-us/copilot/microsoft-365/watermarks",
"evidence_quote": "This policy doesn't apply to images.",
"reason": "Policyen gjelder kun video og lyd; siden sier eksplisitt at den ikke gjelder bilder (bilder er brukerstyrt via myaccount), så den lastbærende delen «bilder ... styrt av policy» er motsagt.",
"file": "skills/ms-ai-security/references/ai-security-engineering/model-fingerprinting-watermarking.md",
"line": 44,
"claim": "Microsoft 365 Copilot kan merke AI-generert innhold — bilder | video | lyd — med watermarks, styrt av policy.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/model-fingerprinting-watermarking.md#9",
"judge_verdict": "not_grounded",
"rule": "R2",
"evidence_url": "https://learn.microsoft.com/en-us/copilot/microsoft-365/watermarks",
"evidence_quote": "You can't customize the placement or the wording of the watermark.",
"reason": "Den kanoniske M365-watermark-siden beskriver kun at plassering og ordlyd ikke kan tilpasses; ingen av de tre påståtte egenskapene (ikke-muterbar, persistent ved printing og screenshots, MIP-labeled PDF-støtte) står der — den matrisen tilhører Microsoft Edge for Business Watermarking Protection, et annet produkt, som lister «Printing and screenshots ✔️» og «MIP-labeled PDFs ✔️».",
"file": "skills/ms-ai-security/references/ai-security-engineering/model-fingerprinting-watermarking.md",
"line": 85,
"claim": "Microsoft 365-watermarks er ikke-muterbare (kan ikke fjernes eller modifiseres av brukeren) | persistente (vises også ved printing og screenshots) | MIP-labeling aware (støtter sensitivity-labeled PDF-er).",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/model-fingerprinting-watermarking.md#10",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/en-us/copilot/microsoft-365/watermarks",
"evidence_quote": "Currently, this additional information is added only to the metadata of images. We're working on getting the information added to video and audio content, but we don't have a specific time frame for when that will happen.",
"reason": "Metadata-delen (modell, app, tidspunkt) og «selv om watermark er av» stemmer, men den lastbærende delen «alle AI-genererte filer» er motsagt: siden sier tilleggsinformasjonen foreløpig kun legges til bilder.",
"file": "skills/ms-ai-security/references/ai-security-engineering/model-fingerprinting-watermarking.md",
"line": 90,
"claim": "Selv om watermark er deaktivert, legges C2PA-metadata (modell, app, timestamp) til i alle AI-genererte filer fra Microsoft 365.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/model-fingerprinting-watermarking.md#11",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/en-us/security/benchmark/azure/mcsb-v2-artificial-intelligence-security",
"evidence_quote": "**Poison Training Data (AML.T0020):** Adversaries introduce malicious data into training datasets to manipulate model behavior, causing the model to produce incorrect outputs, embed backdoors, or leak sensitive information.",
"reason": "MCSB AI-1 lister AML.T0018.000 (Poison AI Model), AML.T0020 (Poison Training Data) og T1195 (Supply Chain Compromise); AML.T0050 finnes ikke der, og AML.T0020 er Poison Training Data — ikke «Compromise Model Supply Chain». Den siterte threat-modeling-siden nevner ingen av ID-ene i det hele tatt.",
"file": "skills/ms-ai-security/references/ai-security-engineering/model-fingerprinting-watermarking.md",
"line": 105,
"claim": "Relevante trusselteknikker er AML.T0050: Backdoor Model | AML.T0020: Compromise Model Supply Chain | T1195: Supply Chain Compromise.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/model-fingerprinting-watermarking.md#19",
"judge_verdict": "not_grounded",
"rule": "R7",
"evidence_url": "https://learn.microsoft.com/en-us/azure/ai-foundry/openai/reference-preview-latest",
"evidence_quote": "POST {endpoint}/openai/v1/images/generations?api-version=preview",
"reason": "Content Credentials-siden oppgir ingen api_version, og den gjeldende REST-referansen for bildegenerering bruker api-version=preview på /openai/v1/images/generations — ikke 2024-05-01-preview; api_version-strengen er selve påstanden og faller under R7s lastbærende unntak.",
"file": "skills/ms-ai-security/references/ai-security-engineering/model-fingerprinting-watermarking.md",
"line": 258,
"claim": "Azure OpenAI-bildegenerering med Content Credentials kalles med api_version \"2024-05-01-preview\".",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/model-fingerprinting-watermarking.md#20",
"judge_verdict": "not_grounded",
"rule": "R2",
"evidence_url": "https://learn.microsoft.com/en-us/azure/foundry-classic/how-to/built-in-policy-model-deployment",
"evidence_quote": "**Allowed Models Publishers**: Enter a list of publisher names enclosed in quotation marks, separated by commas.",
"reason": "Den kanoniske parameter-oppregningen for policyen navngir parameterne Allowed Models Publishers (allowedModelPublishers i Bicep-eksempelet) og allowedAssetIds — allowedPublisherNames finnes ikke, og ingen Learn-side oppgir en defaultValue-liste med Microsoft, OpenAI og Meta; brukeren blir tvert imot bedt om å fylle inn listen selv.",
"file": "skills/ms-ai-security/references/ai-security-engineering/model-fingerprinting-watermarking.md",
"line": 385,
"claim": "Azure Policy-parameteren allowedPublisherNames har defaultValue [\"Microsoft\", \"OpenAI\", \"Meta\"].",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/model-fingerprinting-watermarking.md#21",
"judge_verdict": "source_silent",
"rule": "",
"evidence_url": "https://learn.microsoft.com/en-us/azure/foundry-classic/how-to/built-in-policy-model-deployment",
"evidence_quote": "",
"reason": "Learn oppgir effektene Audit, Deny og Disabled for policyen (allowedValues-delen stemmer), men ingen Learn-side oppgir en defaultValue for effect-parameteren; portalveiledningen ber administratoren sette den til Deny, noe som er en instruksjon og ikke en dokumentert default, så påstanden kan verken bekreftes eller motbevises.",
"file": "skills/ms-ai-security/references/ai-security-engineering/model-fingerprinting-watermarking.md",
"line": 396,
"claim": "Azure Policy-parameteren effect har defaultValue \"Deny\" og allowedValues Audit | Deny | Disabled.",
"disposition": "unsourced"
}
]
},
{
"file": "skills/ms-ai-security/references/ai-security-engineering/norwegian-content-safety.md",
"batch": "R7.5",
"judged_at": "2026-08-01",
"per_file_verdict": "flagged",
"claim_count": 23,
"verified": null,
"verified_by": null,
"flags": [
{
"id": "ms-ai-security/ai-security-engineering/norwegian-content-safety.md#5",
"judge_verdict": "not_grounded",
"rule": "R1",
"evidence_url": "https://learn.microsoft.com/azure/ai-services/content-safety/concepts/harm-categories",
"evidence_quote": "Text: The current version of the text model supports the full 0-7 severity scale. The classifier detects among all severities along this scale. If the user specifies, it can return severities in the trimmed scale of 0, 2, 4, and 6",
"reason": "Påstanden setter taket til 6, men gjeldende tekstmodell bruker hele 0-7-skalaen (0-6 er kun den valgfrie trimmede returskalaen), og ingen side sier at skalaen er konsistent på tvers av språk — språksiden sier tvert imot at kvaliteten kan variere.",
"file": "skills/ms-ai-security/references/ai-security-engineering/norwegian-content-safety.md",
"line": 45,
"claim": "Severity levels i Azure AI Content Safety tekstmoderering går fra 0 til 6, og skalaen er konsistent på tvers av språk.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/norwegian-content-safety.md#17",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/azure/ai-services/content-safety/quickstart-blocklist",
"evidence_quote": "curl --location --request PATCH '<endpoint>/contentsafety/text/blocklists/<your_blocklist_name>?api-version=2024-09-01'",
"reason": "api-version=2024-09-01 og POST :addOrUpdateBlocklistItems stemmer, men den lastbærende HTTP-metoden for opprettelse av blocklist er PATCH i kilden, ikke PUT — én uttalt del som en leser ville kopiert inn i kode er feil.",
"file": "skills/ms-ai-security/references/ai-security-engineering/norwegian-content-safety.md",
"line": 195,
"claim": "Blocklist-API-et for Azure AI Content Safety kalles med api-version=2024-09-01 (PUT /contentsafety/text/blocklists/{name} og POST :addOrUpdateBlocklistItems).",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/norwegian-content-safety.md#18",
"judge_verdict": "source_silent",
"rule": "",
"evidence_url": "https://learn.microsoft.com/azure/ai-services/content-safety/faq",
"evidence_quote": "",
"reason": "FAQ-en nevner kun at bruken strupes ved fri-nivåets transaksjonsgrense uten å tallfeste den; 5K gratis transaksjoner per måned står bare for tekst- og bildemoderering i migreringsguiden, og ingen Learn-side oppgir tallet for Prompt Shields, Protected Material eller Custom Categories (rapid) — selve kvoten ligger på den JS-genererte prissiden.",
"file": "skills/ms-ai-security/references/ai-security-engineering/norwegian-content-safety.md",
"line": 439,
"claim": "Free tier for Azure AI Content Safety gir 5000 transaksjoner per måned for Text Moderation | Image Moderation | Prompt Shields | Protected Material | Custom Categories (rapid); Groundedness Detection er ikke tilgjengelig i free tier.",
"disposition": "unsourced"
},
{
"id": "ms-ai-security/ai-security-engineering/norwegian-content-safety.md#19",
"judge_verdict": "source_silent",
"rule": "",
"evidence_url": "https://learn.microsoft.com/azure/ai-services/translator/service-limits",
"evidence_quote": "",
"reason": "Translator-grensesiden oppgir F0 til 2 millioner tegn per TIME som gjennomstrømningsgrense, ikke en månedskvote; den månedlige gratiskvoten finnes ikke på noen Learn-side (kun på prissiden og i et Q-og-A-svar), så påstanden kan verken bekreftes eller motbevises herfra.",
"file": "skills/ms-ai-security/references/ai-security-engineering/norwegian-content-safety.md",
"line": 445,
"claim": "Azure Translator har en free tier på 2 millioner tegn per måned.",
"disposition": "unsourced"
},
{
"id": "ms-ai-security/ai-security-engineering/norwegian-content-safety.md#23",
"judge_verdict": "not_grounded",
"rule": "R2",
"evidence_url": "https://learn.microsoft.com/azure/ai-services/content-safety/overview",
"evidence_quote": "France Central | Germany West Central | Italy North | Poland Central | Sweden Central | Switzerland North | Switzerland West | UK South | West Europe",
"reason": "Den kanoniske regionstabellen for Content Safety lister West Europe, men Norway East forekommer ikke i det hele tatt — fraværet i den uttømmende opplistingen gjør at påstanden om Norway East-deploy er feil, ikke bare ubekreftet.",
"file": "skills/ms-ai-security/references/ai-security-engineering/norwegian-content-safety.md",
"line": 496,
"claim": "Azure AI Content Safety kan deployes i regionene Norway East eller West Europe for data residency i EU/EØS.",
"disposition": "outdated"
}
]
},
{
"file": "skills/ms-ai-security/references/ai-security-engineering/output-validation-grounding-verification.md",
"batch": "R7.5",
"judged_at": "2026-08-01",
"per_file_verdict": "flagged",
"claim_count": 19,
"verified": null,
"verified_by": null,
"flags": [
{
"id": "ms-ai-security/ai-security-engineering/output-validation-grounding-verification.md#1",
"judge_verdict": "not_grounded",
"rule": "R2",
"evidence_url": "https://learn.microsoft.com/azure/ai-services/content-safety/overview",
"evidence_quote": "Groundedness detection (preview) | Detects whether the text responses of large language models (LLMs) are grounded in the source materials provided by the users.",
"reason": "Oversiktssidens funksjonstabell og rate-tabell merker funksjonen «(preview)», og quickstarten heter «Quickstart: Use Groundedness detection (preview)» — ingen GA-status finnes på den kanoniske siden, så påstanden om GA er motsagt.",
"file": "skills/ms-ai-security/references/ai-security-engineering/output-validation-grounding-verification.md",
"line": 4,
"claim": "Groundedness-deteksjon i Azure AI Content Safety oppgis med status GA (generelt tilgjengelig).",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/output-validation-grounding-verification.md#15",
"judge_verdict": "source_silent",
"rule": "",
"evidence_url": "https://learn.microsoft.com/ai-builder/use-in-flow-overview",
"evidence_quote": "",
"reason": "AI Builder-kapabilitetslisten nevner ikke groundedness, men ingen Learn-side sier at AI Builder mangler et native groundedness-API eller at det må integreres via custom connector til Azure AI Content Safety eller en Power Automate-flow — den lastbærende integrasjonsanvisningen er udokumentert.",
"file": "skills/ms-ai-security/references/ai-security-engineering/output-validation-grounding-verification.md",
"line": 429,
"claim": "Power Platform AI Builder har ingen native groundedness-API; groundedness må integreres via custom connector til Azure AI Content Safety eller Power Automate-flow.",
"disposition": "unsourced"
},
{
"id": "ms-ai-security/ai-security-engineering/output-validation-grounding-verification.md#16",
"judge_verdict": "not_grounded",
"rule": "R2",
"evidence_url": "https://learn.microsoft.com/azure/ai-services/commitment-tier",
"evidence_quote": "With commitment tier pricing, you can commit to using the following Foundry Tools features for a fixed fee ... Speech to text (Standard); Text to speech (Neural); Text Translation (Standard); Language Understanding standard (Text Requests); Azure Language ...; Vision in Foundry Tools - OCR; Document Intelligence – Custom/Invoice",
"reason": "Den kanoniske commitment-tier-siden enumererer hvilke tjenester som kan kjøpes på commitment, og verken Content Safety eller Groundedness står der — pay-as-you-go-delen (S0) stemmer, men commitment-delen er motsagt av enumerasjonen.",
"file": "skills/ms-ai-security/references/ai-security-engineering/output-validation-grounding-verification.md",
"line": 513,
"claim": "Groundedness API er inkludert i Azure AI Services commitment (Foundry-lisenser) og tilbys consumption-based (pay-as-you-go).",
"disposition": "outdated"
}
]
},
{
"file": "skills/ms-ai-security/references/ai-security-engineering/owasp-llm-top10-azure-mitigations.md",
"batch": "R7.5",
"judged_at": "2026-08-01",
"per_file_verdict": "flagged",
"claim_count": 18,
"verified": null,
"verified_by": null,
"flags": [
{
"id": "ms-ai-security/ai-security-engineering/owasp-llm-top10-azure-mitigations.md#5",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/microsoft-copilot-studio/guidance/autonomous-agents",
"evidence_quote": "Autonomous agents in Copilot Studio extend the value of generative orchestration by enabling AI to take action without waiting for a user prompt. ... Implement human oversight for critical actions: For high-stakes tasks, keep a human in the loop. Configure the agent to request approval or confirmation from a person before executing actions that could be sensitive.",
"reason": "Første del holder (Agent Framework bruker @tool(approval_mode=\"always_require\")), men den lastbærende andre delen er motsagt: kilden beskriver autonome agenter som handler uten brukerprompt og gjør godkjenning til noe du selv må konfigurere for kritiske handlinger, ikke en garanti om at Copilot Studio aldri skriver til Dataverse uten brukergodkjenning.",
"file": "skills/ms-ai-security/references/ai-security-engineering/owasp-llm-top10-azure-mitigations.md",
"line": 45,
"claim": "Microsoft Agent Framework støtter human-in-the-loop via innstillingen approval_mode: always_require, og Copilot Studio skriver aldri til Dataverse uten brukergodkjenning.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/owasp-llm-top10-azure-mitigations.md#16",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/azure/foundry/concepts/evaluation-evaluators/rag-evaluators",
"evidence_quote": "Response Completeness (preview) | System evaluation | You want to ensure the RAG response doesn't miss critical information (recall aspect) from your ground truth | Measures how completely the response covers the expected information compared to ground truth",
"reason": "Groundedness-evaluatoren er på GA-nivå og Foundry Evaluations er GA, men completeness-metrikken er eksplisitt merket Response Completeness (preview), så den lastbærende GA-påstanden om metrikk-paret er motsagt av kilden.",
"file": "skills/ms-ai-security/references/ai-security-engineering/owasp-llm-top10-azure-mitigations.md",
"line": 79,
"claim": "AI Foundry Evaluation med metrikkene groundedness og completeness for end-to-end LLM-evaluering er GA.",
"disposition": "outdated"
}
]
},
{
"file": "skills/ms-ai-security/references/ai-security-engineering/pii-detection-norwegian-context.md",
"batch": "R7.5",
"judged_at": "2026-08-01",
"per_file_verdict": "flagged",
"claim_count": 19,
"verified": null,
"verified_by": null,
"flags": [
{
"id": "ms-ai-security/ai-security-engineering/pii-detection-norwegian-context.md#8",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/en-us/azure/ai-services/language-service/whats-new",
"evidence_quote": "Use entity synonyms for tailored PII detection.",
"reason": "What's new plasserer entitySynonyms og valueExclusionPolicy under mai 2025 (2025-05-15-preview), mens 2025-11-15-preview kun introduserte syntheticReplacement, disableEntityValidation og confidenceScoreThreshold — to lastbærende deler av påstanden er dermed feil.",
"file": "skills/ms-ai-security/references/ai-security-engineering/pii-detection-norwegian-context.md",
"line": 67,
"claim": "Preview-funksjonene redaction-policies | `confidenceScoreThreshold` | `disableEntityValidation` | `entitySynonyms` | `valueExclusionPolicy` ble først introdusert i API-versjon `2025-11-15-preview` og er dokumentert under gjeldende preview-versjon.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/pii-detection-norwegian-context.md#16",
"judge_verdict": "not_grounded",
"rule": "R1",
"evidence_url": "https://learn.microsoft.com/en-us/azure/ai-services/language-service/concepts/data-limits",
"evidence_quote": "Personally Identifying Information (PII) detection | 5",
"reason": "Tjenestegrensene setter maks dokumenter per request for PII-deteksjon til 5 (maks 25 ved asynkrone kall); taket på 5000 finnes ikke — høyeste verdi i hele tabellen er 1000 for språkdeteksjon.",
"file": "skills/ms-ai-security/references/ai-security-engineering/pii-detection-norwegian-context.md",
"line": 262,
"claim": "Azure AI Language støtter batch processing med opptil 5000 dokumenter per request.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/pii-detection-norwegian-context.md#18",
"judge_verdict": "source_silent",
"rule": "",
"evidence_url": "https://learn.microsoft.com/en-us/azure/ai-services/containers/container-faq",
"evidence_quote": "",
"reason": "Ingen learn.microsoft.com-side oppgir en SLA-prosent for Azure Language; container-FAQ-en peker videre til den eksterne SLA-siden, så 99,9 % kan verken bekreftes eller avkreftes fra Learn.",
"file": "skills/ms-ai-security/references/ai-security-engineering/pii-detection-norwegian-context.md",
"line": 340,
"claim": "Standard-tieren (S) for Azure AI Language inkluderer alle features og har SLA 99,9 %.",
"disposition": "unsourced"
},
{
"id": "ms-ai-security/ai-security-engineering/pii-detection-norwegian-context.md#19",
"judge_verdict": "not_grounded",
"rule": "",
"evidence_url": "https://learn.microsoft.com/en-us/azure/ai-services/language-service/concepts/data-limits",
"evidence_quote": "Pricing is based on the number of text records you send to the API, and is subject to your Language resource's pricing details. **A text record is measured as 1000 characters**.",
"reason": "Kilden sier at ett text record måles som 1000 tegn; 5 120 er maks antall tegn per dokument for synkrone kall, altså en annen størrelse enn påstanden hevder.",
"file": "skills/ms-ai-security/references/ai-security-engineering/pii-detection-norwegian-context.md",
"line": 343,
"claim": "Ett text record i Azure AI Language tilsvarer opptil 5120 tegn.",
"disposition": "outdated"
}
]
},
{
"file": "skills/ms-ai-security/references/ai-security-engineering/prompt-injection-defense-patterns.md",
"batch": "R7.5",
"judged_at": "2026-08-01",
"per_file_verdict": "flagged",
"claim_count": 14,
"verified": null,
"verified_by": null,
"flags": [
{
"id": "ms-ai-security/ai-security-engineering/prompt-injection-defense-patterns.md#1",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/en-us/azure/ai-services/content-safety/concepts/jailbreak-detection",
"evidence_quote": "Encoding Attacks | This attack attempts to use encoding, such as a character transformation method, generation styles, ciphers, or other natural language variations, to circumvent the system rules.",
"reason": "Siden sin tabell «Types of input attacks» lister BARE to hovedtyper (User Prompt attacks og Document attacks), og Encoding Attacks er en undertype under begge — påstandens lastbærende del om en tredje hovedtype «Encoding-basert» er derfor motsagt av kilden.",
"file": "skills/ms-ai-security/references/ai-security-engineering/prompt-injection-defense-patterns.md",
"line": 26,
"claim": "Prompt injection-angrep deles i tre hovedtyper: Direkte (jailbreaking) | Indirekte (ondsinnet innhold skjult i eksterne dokumenter/data) | Encoding-basert (koding, transformasjoner eller språkvarianter for å omgå filtre).",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/prompt-injection-defense-patterns.md#7",
"judge_verdict": "not_grounded",
"rule": "R2",
"evidence_url": "https://learn.microsoft.com/en-us/azure/role-based-access-control/built-in-roles/ai-machine-learning",
"evidence_quote": "\"roleName\": \"Cognitive Services OpenAI User\"",
"reason": "Den kanoniske opplistingen av innebygde AI-roller inneholder ingen rolle som heter «Azure AI Services OpenAI User» — rollenavnet er «Cognitive Services OpenAI User», og rollenavnet er nøyaktig det en leser ville kopiert inn i en rolletildeling.",
"file": "skills/ms-ai-security/references/ai-security-engineering/prompt-injection-defense-patterns.md",
"line": 189,
"claim": "Azure-rollen «Azure AI Services OpenAI User» tildeles en Managed Identity (SystemAssigned) med scope mot en Microsoft.CognitiveServices/accounts-ressurs for minste-privilegium-tilgang.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/prompt-injection-defense-patterns.md#11",
"judge_verdict": "not_grounded",
"rule": "R2",
"evidence_url": "https://learn.microsoft.com/python/api/azure-ai-contentsafety/azure.ai.contentsafety.contentsafetyclient?view=azure-python",
"evidence_quote": "A synchronous API for the analysis of potentially harmful text content. Currently, it supports four categories: Hate, SelfHarm, Sexual, and Violence.",
"reason": "Den kanoniske SDK-opplistingen av TextCategory har ingen «Jailbreak»-kategori, og responsen eksponerer categoriesAnalysis — ikke jailbreak_analysis.attack_detected; kun output_type «FourSeverityLevels» stemmer, og jailbreak-deteksjon går via det separate shieldPrompt-kallet.",
"file": "skills/ms-ai-security/references/ai-security-engineering/prompt-injection-defense-patterns.md",
"line": 302,
"claim": "ContentSafetyClient.analyze_text støtter kategorien «Jailbreak» og output_type «FourSeverityLevels», og resultatet eksponerer jailbreak_analysis.attack_detected.",
"disposition": "outdated"
}
]
},
{
"file": "skills/ms-ai-security/references/ai-security-engineering/security-copilot-integration.md",
"batch": "R7.5",
"judged_at": "2026-08-01",
"per_file_verdict": "flagged",
"claim_count": 34,
"verified": null,
"verified_by": null,
"flags": [
{
"id": "ms-ai-security/ai-security-engineering/security-copilot-integration.md#1",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/copilot/security/security-copilot-inclusion",
"evidence_quote": "Eligible Microsoft 365 E5 and E7 customers who were notified of their eligibility will automatically receive Security Copilot through zero click activation. For eligible customers, Security Copilot is provisioned automatically as part of their subscription. This means no Azure setup is needed or capacity provisioning required.",
"reason": "Inklusjon i E5/E7 og auto-provisjonering uten SCU-kjøp stemmer, men den lastbærende delen om en 7-dagers forhåndsvarsling finnes ikke på noen Learn-side (auto-provisioning-siden lister hva som provisjoneres uten noe varslingsvindu, og eneste dokumenterte frist er 30 dager for fremtidig pay-as-you-go).",
"file": "skills/ms-ai-security/references/ai-security-engineering/security-copilot-integration.md",
"line": 37,
"claim": "Security Copilot er inkludert i lisensen for Microsoft 365 E5/E7 inclusion-kunder og auto-provisjoneres etter en 7-dagers forhåndsvarsling fra Microsoft; ingen SCU-kjøp er nødvendig for grunnfunksjonalitet, og aktivering skjer per tenant.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/security-copilot-integration.md#4",
"judge_verdict": "not_grounded",
"rule": "R4",
"evidence_url": "https://learn.microsoft.com/defender-xdr/security-alert-triage-agent",
"evidence_quote": "Email and collaboration alert triage capabilities are already generally available (GA).",
"reason": "Public preview-statusen stammer fra en tidsstemplet what's-new-post fra juli 2025; gjeldende dokumentasjon sier at phishing-/e-posttriagen er GA, og bare de utvidede sky- og identitetsalarmene er i preview.",
"file": "skills/ms-ai-security/references/ai-security-engineering/security-copilot-integration.md",
"line": 83,
"claim": "Phishing Triage Agent i Defender XDR er i Public Preview.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/security-copilot-integration.md#7",
"judge_verdict": "not_grounded",
"rule": "R4",
"evidence_url": "https://learn.microsoft.com/defender-xdr/security-copilot-agents-defender",
"evidence_quote": "1. Security Alert Triage Agent (Preview) 2. Threat Intelligence Briefing Agent 3. Threat Hunting Agent 4. Security Analyst Agent 5. Dynamic Threat Detection Agent 6. Data Security Triage Agent in Data Loss Prevention",
"reason": "«Public Preview» står kun i en tidsstemplet what's-new-post fra juli 2025; de gjeldende kanoniske sidene merker preview-agenter eksplisitt, og Threat Intelligence Briefing Agent står umerket både her og i standalone-artikkelen, som heller ikke har noe preview-varsel.",
"file": "skills/ms-ai-security/references/ai-security-engineering/security-copilot-integration.md",
"line": 91,
"claim": "Threat Intelligence Briefing Agent i standalone-portalen er i Public Preview.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/security-copilot-integration.md#9",
"judge_verdict": "not_grounded",
"rule": "",
"evidence_url": "https://learn.microsoft.com/intune/copilot/agents/vulnerability-remediation-agent",
"evidence_quote": "This feature is in public preview. For more information, see Public preview in Microsoft Intune.",
"reason": "Agentsiden merker Vulnerability Remediation Agent som public preview (og admin-senteret viser «Vulnerability Remediation Agent (preview)»), mens påstanden sier GA.",
"file": "skills/ms-ai-security/references/ai-security-engineering/security-copilot-integration.md",
"line": 93,
"claim": "Vulnerability Remediation Agent i Microsoft Intune er GA.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/security-copilot-integration.md#10",
"judge_verdict": "not_grounded",
"rule": "R2",
"evidence_url": "https://learn.microsoft.com/entra/security-copilot/entra-agents",
"evidence_quote": "The following agents are currently available for Microsoft Entra. ... Conditional Access Optimization Agent ... Identity Risk Management Agent (Preview)",
"reason": "Access Review Agent er fraværende i den kanoniske Entra-agentlisten, og eneste statusangivelse på Learn er what's-new-posten «Microsoft and partner agents - Public preview» med Access Review Agent i Microsoft Entra — ingen kilde sier GA.",
"file": "skills/ms-ai-security/references/ai-security-engineering/security-copilot-integration.md",
"line": 94,
"claim": "Access Review Agent i Microsoft Entra + Teams er GA.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/security-copilot-integration.md#11",
"judge_verdict": "not_grounded",
"rule": "",
"evidence_url": "https://learn.microsoft.com/intune/agents/",
"evidence_quote": "Microsoft Intune includes specialized Security Copilot agents, each designed for a specific security scenario. The following agents are available: Change Review Agent ... Device Offboarding Agent ... Policy Configuration Agent ... Vulnerability Remediation Agent",
"reason": "Påstanden sier at agentene «består av» tre navngitte agenter, mens den kanoniske Intune-siden lister fire — Vulnerability Remediation Agent mangler i den lukkede oppregningen.",
"file": "skills/ms-ai-security/references/ai-security-engineering/security-copilot-integration.md",
"line": 100,
"claim": "Security Copilot-agentene for endepunktadministrasjon i Intune består av: Change Review Agent | Device Offboarding Agent | Policy Configuration Agent.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/security-copilot-integration.md#17",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/copilot/security/security-copilot-inclusion",
"evidence_quote": "Customers will get a 30-day advanced notification when this option is available.",
"reason": "Zero-click-delen stemmer, men den lastbærende koblingen mellom et 30-dagers forhåndsvarsel og auto-provisjoneringen er feil — kilden knytter 30-dagersvarselet til den fremtidige pay-as-you-go-opsjonen, ikke til provisjoneringen.",
"file": "skills/ms-ai-security/references/ai-security-engineering/security-copilot-integration.md",
"line": 120,
"claim": "Kunder mottar 30-dagers forhåndsvarsel, deretter auto-provisjoneres Security Copilot uten Azure-oppsett eller manuell SCU-tildeling (zero-click activation).",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/security-copilot-integration.md#27",
"judge_verdict": "not_grounded",
"rule": "",
"evidence_url": "https://learn.microsoft.com/copilot/security/plugin-overview",
"evidence_quote": "A plugin is a collection of related tools that users can enable in Security Copilot when relevant to creating new prompting, promptbook, or agentic capabilities. ... For agents, plugins extend what an agent can do by giving it access to resources outside of an agent and LLM.",
"reason": "Learn kategoriserer egendefinerte plugins som API-, KQL- og GPT-plugins (egne artikler plugin-api, plugin-kql, plugin-gpt) pluss MCP, mens påstandens lukkede liste mangler GPT-plugin og fører opp «egendefinert agent», som ikke er en plugintype — plugins er noe som utvider agenter.",
"file": "skills/ms-ai-security/references/ai-security-engineering/security-copilot-integration.md",
"line": 235,
"claim": "Tilpassede Security Copilot-plugins finnes i typene: API-plugin (OpenAPI-spec-wrapper rundt REST API) | KQL-plugin (egendefinerte KQL-spørringer mot Sentinel/Defender) | OpenAI-format (ChatGPT-kompatibelt plugin-format) | Egendefinert agent.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/security-copilot-integration.md#33",
"judge_verdict": "not_grounded",
"rule": "R2",
"evidence_url": "https://learn.microsoft.com/copilot/security/plugin-other",
"evidence_quote": "AbuseIPDB ... Aviatrix ... Censys ... CheckPhish ... CIRCL Hash Lookup ... CrowdSec ... CyberArk Privilege Cloud ... Cybersixgill ... Cyware Intel Exchange ... Darktrace ... Jamf ... Netskope ... Quest Security Guardian ... ReversingLabs ... Saviynt ... ServiceNow SIR ... Shodan ... Splunk ... Tanium ... UrlScan",
"reason": "Den kanoniske oversikten over ikke-Microsoft-plugins inneholder seks av de syv navnene, men Red Canary er ikke oppført i det hele tatt.",
"file": "skills/ms-ai-security/references/ai-security-engineering/security-copilot-integration.md",
"line": 272,
"claim": "Tredjepartspluginer tilgjengelige via Security Store inkluderer: AbuseIPDB | Censys | CrowdSec CTI | CyberArk | Cybersixgill | Red Canary | Jamf.",
"disposition": "outdated"
}
]
},
{
"file": "skills/ms-ai-security/references/ai-security-engineering/security-scoring-rubrics-6x5.md",
"batch": "R7.5",
"judged_at": "2026-08-01",
"per_file_verdict": "flagged",
"claim_count": 21,
"verified": null,
"verified_by": null,
"flags": [
{
"id": "ms-ai-security/ai-security-engineering/security-scoring-rubrics-6x5.md#7",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/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**|",
"reason": "Norway East holder, men norwaywest mangler helt i Europa-tabellen for Foundry-modeller og i Azure AI Search sin regionliste, og regionoversikten merker Norway West som en tilgangsbegrenset region for katastrofegjenoppretting — den lastbærende delen om provisjonering av AI-ressurser i Norway West er ikke dekket.",
"file": "skills/ms-ai-security/references/ai-security-engineering/security-scoring-rubrics-6x5.md",
"line": 144,
"claim": "Azure-regionene Norway East (`norwayeast`) og Norway West (`norwaywest`) er tilgjengelige for provisjonering av AI-ressurser med data residency i Norge.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/security-scoring-rubrics-6x5.md#12",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/azure/defender-for-cloud/ai-onboarding",
"evidence_quote": "On the Defender plans page, toggle the AI services to **On**.",
"reason": "AI security posture management ligger riktignok i Defender CSPM-planen, men AI threat protection aktiveres i en egen AI services-plan — den lastbærende koblingen av trusselbeskyttelse til Defender CSPM-planen er feil.",
"file": "skills/ms-ai-security/references/ai-security-engineering/security-scoring-rubrics-6x5.md",
"line": 221,
"claim": "AI security posture management (AI SPM) / AI threat protection leveres i Defender for Cloud via Defender CSPM-planen, aktivert under Environment Settings.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/security-scoring-rubrics-6x5.md#14",
"judge_verdict": "not_grounded",
"rule": "",
"evidence_url": "https://learn.microsoft.com/microsoft-copilot-studio/admin-network-isolation-vnet",
"evidence_quote": "Copilot Studio integrates with Power Platform virtual networks over a private endpoint for these scenarios:",
"reason": "Kilden viser at Copilot Studio faktisk integreres over private endpoints via Power Platform Virtual Network-støtte (Key Vault over HTTP, Application Insights, VNet-støttede koblinger), så påstanden om at private endpoints ikke er mulig er motsagt.",
"file": "skills/ms-ai-security/references/ai-security-engineering/security-scoring-rubrics-6x5.md",
"line": 277,
"claim": "Copilot Studio er en SaaS-tjeneste der private endpoints ikke er mulig, og NSG-regler ikke er relevante.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/security-scoring-rubrics-6x5.md#19",
"judge_verdict": "not_grounded",
"rule": "R2",
"evidence_url": "https://learn.microsoft.com/security/benchmark/azure/overview",
"evidence_quote": "Network security (NS) ... Identity Management (IM) ... Privileged Access (PA) ... Data Protection (DP) ... Asset Management (AM) ... Logging and Threat Detection (LT) ... Incident Response (IR) ... Posture and Vulnerability Management (PV) ... Endpoint Security (ES) ... Backup and Recovery (BR) ... DevOps Security (DS) ... Artificial Intelligence Security (AI)",
"reason": "Den kanoniske domenetabellen for MCSB v2 lister tolv domener, og Governance and Strategy (GS) er ikke blant dem — GS var et v1-domene; i tillegg heter loggekontrollen i v2 LT-3, ikke LT-4 slik påstanden oppgir.",
"file": "skills/ms-ai-security/references/ai-security-engineering/security-scoring-rubrics-6x5.md",
"line": 343,
"claim": "MCSB v2 er inndelt i security domains som omfatter IM (Identity Management) | PA (Privileged Access) | NS (Network Security) | DP (Data Protection) | AI (Artificial Intelligence Security) | GS (Governance and Strategy) | LT (Logging and Threat Detection) | IR (Incident Response), med kontroller som IM-1 | IM-3 | IM-7 | IM-8 | PA-1 | PA-7 | NS-1 | NS-2 | DP-1 til DP-6 | LT-1 | LT-4.",
"disposition": "outdated"
}
]
},
{
"file": "skills/ms-ai-security/references/ai-security-engineering/supply-chain-security-ai-models.md",
"batch": "R7.5",
"judged_at": "2026-08-01",
"per_file_verdict": "flagged",
"claim_count": 17,
"verified": null,
"verified_by": null,
"flags": [
{
"id": "ms-ai-security/ai-security-engineering/supply-chain-security-ai-models.md#4",
"judge_verdict": "not_grounded",
"rule": "R2",
"evidence_url": "https://learn.microsoft.com/azure/devops/pipelines/tasks/reference/advanced-security-dependency-scanning-v1?view=azure-pipelines",
"evidence_quote": "Inputs: `directoryExclusionList` - Directory exclusion list `string`. List of relative directory paths to ignore as a set of semi-colon separated values.",
"reason": "Oppgavenavnet og versjon 1 stemmer, men den kanoniske oppgavereferansen enumererer kun inputen directoryExclusionList — verken scanMode eller ecosystem finnes i input-listen, og fraværet i den autoritative enumereringen er bevis mot påstanden.",
"file": "skills/ms-ai-security/references/ai-security-engineering/supply-chain-security-ai-models.md",
"line": 126,
"claim": "Azure Pipelines-oppgaven for avhengighetsskanning heter AdvancedSecurity-Dependency-Scanning@1 (versjon 1) og tar inputene scanMode og ecosystem (f.eks. «pip»).",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/supply-chain-security-ai-models.md#5",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/azure/devops/repos/security/github-advanced-security-dependency-scanning?view=azure-devops",
"evidence_quote": "Dependency scanning generates an alert for any open-source component, direct or transitive, found to be vulnerable that your code depends upon.",
"reason": "Kilden beskriver to alert-kategorier (direkte og transitive sårbare komponenter); CVSS-til-alvorlighet er en egenskap ved en alert, ikke en tredje alert-kategori, så den lastbærende «tre kategorier»-delen er feil.",
"file": "skills/ms-ai-security/references/ai-security-engineering/supply-chain-security-ai-models.md",
"line": 134,
"claim": "Dependency scanning genererer alerts for tre kategorier: Direct vulnerabilities (pakker i requirements.txt) | Transitive vulnerabilities (pakker som direkte dependencies bruker) | CVE severity mapping.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/supply-chain-security-ai-models.md#8",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/azure/defender-for-cloud/runtime-gated-overview",
"evidence_quote": "Gated deployment is a Microsoft Defender for Containers capability that uses an admission controller to evaluate container images before they're admitted into a Kubernetes cluster. ... Deny blocks deployments that match the rule conditions.",
"reason": "Skanning ved push stemmer, men blokkering av images med kritiske sårbarheter konfigureres via Defender for Containers' gated deployment-sikkerhetsregler i Defender for Cloud, ikke via Azure Policy — den relevante Azure Policy-definisjonen for sårbare running images har kun effektene AuditIfNotExists og Disabled, så den lastbærende Azure Policy-delen er feil.",
"file": "skills/ms-ai-security/references/ai-security-engineering/supply-chain-security-ai-models.md",
"line": 157,
"claim": "Microsoft Defender for Containers gjør tre ting: genererer vulnerability assessments automatisk når et image pushes til Azure Container Registry | blokkerer deployment av images med kritiske sårbarheter (konfigurerbart via Azure Policy) | integrerer med Azure Monitor for alerting.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/supply-chain-security-ai-models.md#9",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/azure/machine-learning/how-to-deploy-models-from-huggingface?view=azureml-api-2#frequently-asked-questions",
"evidence_quote": "`HuggingFace` is a community registry and Microsoft support doesn't cover it. ... The model weights aren't hosted on Azure.",
"reason": "HuggingFace-registeret beskrives som et community registry som Microsoft-support ikke dekker, og modellvektene lastes ned direkte fra Hugging Face hub — det er derfor ikke en Microsoft-kanal for verifiserte modeller med provenance tracking, så den ene lastbærende halvdelen av påstanden er feil.",
"file": "skills/ms-ai-security/references/ai-security-engineering/supply-chain-security-ai-models.md",
"line": 202,
"claim": "Microsoft tilbyr verifiserte modeller via to kanaler: Azure Machine Learning Model Catalog (kuraterte modeller med security attestation) | HuggingFace Registry i Azure (integrert med Azure ML, med provenance tracking).",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/supply-chain-security-ai-models.md#10",
"judge_verdict": "not_grounded",
"rule": "R4",
"evidence_url": "https://learn.microsoft.com/azure/ai-foundry/openai/concepts/legacy-models",
"evidence_quote": "These models are no longer available for new deployments. ... `gpt-35-turbo` - 0301 ... Retirement date: February 13, 2025",
"reason": "gpt-35-turbo versjon 0301 står i tabellen over pensjonerte modeller med pensjoneringsdato 13. februar 2025 og er ikke lenger tilgjengelig for nye deployeringer; i tillegg unntar Learn eksplisitt Azure OpenAI-modeller fra registry-modell-ID-mønsteret påstanden bruker, så påstanden treffer kun en utdatert rad.",
"file": "skills/ms-ai-security/references/ai-security-engineering/supply-chain-security-ai-models.md",
"line": 209,
"claim": "Modellen gpt-35-turbo finnes i azureml-registeret med modellversjon 0301, referert som azureml://registries/azureml/models/gpt-35-turbo/versions/0301.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/supply-chain-security-ai-models.md#12",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/azure/foundry-classic/how-to/built-in-policy-model-deployment#enable-the-policy",
"evidence_quote": "Effect: Set to Deny. ... Allowed Models Publishers: Enter a list of publisher names enclosed in quotation marks, separated by commas. ... Allowed Asset Ids: Enter a list of model asset IDs enclosed in quotation marks, separated by commas.",
"reason": "Policyen finnes i preview (1.0.0-preview) og støtter Deny, men de dokumenterte parameterne heter «Allowed Models Publishers» og «Allowed Asset Ids» (allowedPublishers/allowedAssetIds i CLI-eksempelet) — approvedAssetIds forekommer ingen steder, og det er en lastbærende streng en leser ville kopiert inn i en parameterfil.",
"file": "skills/ms-ai-security/references/ai-security-engineering/supply-chain-security-ai-models.md",
"line": 255,
"claim": "Azure Policy-definisjonen «[Preview]: Azure Machine Learning Deployments should only use approved Registry Models» er i preview og støtter effect «Deny» med parameterne allowedPublishers og approvedAssetIds.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/supply-chain-security-ai-models.md#13",
"judge_verdict": "source_silent",
"rule": "",
"evidence_url": "https://learn.microsoft.com/azure/machine-learning/concept-online-deployment-model-specification?view=azureml-api-2",
"evidence_quote": "",
"reason": "Registeret azureml-meta og ID-mønsteret azureml://registries/{registry}/models/{model}/versions/{version} er dokumentert, men ingen learn.microsoft.com-side oppgir et versjonsnummer for Llama-2-7b; versjonsnumre for registermodeller er verdier Learn ikke enumererer, så versjon 18 kan verken bekreftes eller avkreftes.",
"file": "skills/ms-ai-security/references/ai-security-engineering/supply-chain-security-ai-models.md",
"line": 260,
"claim": "Modellen Llama-2-7b finnes i registeret azureml-meta med versjon 18, referert som azureml://registries/azureml-meta/models/Llama-2-7b/versions/18.",
"disposition": "unsourced"
},
{
"id": "ms-ai-security/ai-security-engineering/supply-chain-security-ai-models.md#14",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/azure/ai-services/anomaly-detector/overview",
"evidence_quote": "Starting 20 September 2023 you won't be able to create new Anomaly Detector resources. The Anomaly Detector service is being retired 1 October 2026.",
"reason": "Kilden sier at nye Anomaly Detector-ressurser ikke kan opprettes siden 20. september 2023 og at tjenesten pensjoneres 1. oktober 2026, så den lastbærende delen om at tjenesten er tilgjengelig og kan deployes, er direkte motsagt.",
"file": "skills/ms-ai-security/references/ai-security-engineering/supply-chain-security-ai-models.md",
"line": 276,
"claim": "Azure AI Anomaly Detector er tilgjengelig som tjeneste og kan deployes for å identifisere data poisoning i treningsdata (klienten AnomalyDetectorClient med detect_entire_series).",
"disposition": "outdated"
},
{
"id": "ms-ai-security/ai-security-engineering/supply-chain-security-ai-models.md#15",
"judge_verdict": "not_grounded",
"rule": "R2",
"evidence_url": "https://learn.microsoft.com/azure/machine-learning/reference-yaml-environment?view=azureml-api-2",
"evidence_quote": "YAML syntax: `$schema`, `name`, `version`, `description`, `tags`, `image`, `conda_file`, `build`, `build.path`, `build.dockerfile_path`, `os_type`, `inference_config`",
"reason": "Det kanoniske environment-skjemaet enumererer alle felt en environment-definisjon bærer, og ingen kryptografisk hash av miljødefinisjonen finnes blant dem; fraværet i den autoritative enumereringen gjør den lastbærende SBOM-mekanismen ugrunnet.",
"file": "skills/ms-ai-security/references/ai-security-engineering/supply-chain-security-ai-models.md",
"line": 345,
"claim": "Azure ML gir delvis SBOM-funksjonalitet via tre mekanismer: Model Registry Metadata (navn, versjon, tags, properties, koblet treningsjobb) | Environment Registry (conda-avhengigheter, pip-pakker, Docker base image, kryptografisk hash av miljødefinisjonen) | Dataset Versioning (Azure ML Data Assets med versjonering og lineage tracking).",
"disposition": "outdated"
}
]
},
{
"file": "skills/ms-ai-security/references/ai-security-engineering/zero-trust-ai-services.md",
"batch": "R7.1",
@ -11629,6 +12280,77 @@
}
]
},
{
"file": "skills/ms-ai-security/references/cost-optimization/ai-builder-credits-transition.md",
"batch": "R7.5",
"judged_at": "2026-08-01",
"per_file_verdict": "flagged",
"claim_count": 23,
"verified": null,
"verified_by": null,
"flags": [
{
"id": "ms-ai-security/cost-optimization/ai-builder-credits-transition.md#4",
"judge_verdict": "not_grounded",
"rule": "R1",
"evidence_url": "https://learn.microsoft.com/en-us/ai-builder/credit-management",
"evidence_quote": "Power Apps per app | 250 ... Dynamics 365 F&O | 20,000 ... Power Apps for Cloud for Sustainability USL Plus | 25,000 AI Builder credits",
"reason": "Fjerningsdatoen 1. november 2026 stemmer, men den oppgitte øvre grensen i spennet er feil: kanonisk tabell viser 25 000 som høyeste seeded-verdi, ikke 20 000, så taket påstanden setter er overskredet.",
"file": "skills/ms-ai-security/references/cost-optimization/ai-builder-credits-transition.md",
"line": 40,
"claim": "Seeded AI Builder credits som er inkludert i premium-lisenser (varierer fra 250 til 20 000 per måned) fjernes 1. november 2026.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/cost-optimization/ai-builder-credits-transition.md#7",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/en-us/microsoft-copilot-studio/requirements-messages-management",
"evidence_quote": "These rates apply to all language models that Copilot Studio provides. They exclude bring-your-own-model configurations, including Azure Foundry models, which are billed separately.",
"reason": "Copilot Studio-, AI Builder- og Microsoft 365 Copilot-delene holder, men Foundry-delen er lastbærende og motsies: Foundry-forbruk måles og faktureres separat, og administer-licensing definerer Copilot Credits som felles valuta på tvers av Copilot Studio-kapabiliteter — ikke Foundry.",
"file": "skills/ms-ai-security/references/cost-optimization/ai-builder-credits-transition.md",
"line": 57,
"claim": "Copilot Credits er en felles valuta for AI-kapasitet på tvers av Copilot Studio | AI Builder | Microsoft 365 Copilot | Microsoft Foundry.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/cost-optimization/ai-builder-credits-transition.md#9",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/en-us/microsoft-copilot-studio/billing-licensing",
"evidence_quote": "Copilot Credits are the common currency across Copilot Studio capabilities. You can get them through pay-as-you-go meters, prepurchase plans, and Copilot Credit prepaid pack subscriptions.",
"reason": "Beskrivelsen av de to nevnte modellene stemmer, men antallet er lastbærende og feil: kilden lister tre kjøpsveier — pay-as-you-go-målere, prepurchase-planer (CCCU i Azure-portalen) og prepaid pack-abonnement.",
"file": "skills/ms-ai-security/references/cost-optimization/ai-builder-credits-transition.md",
"line": 63,
"claim": "Copilot Credits kan kjøpes gjennom to modeller: prepaid pack subscription (månedlig kapasitetspakke) | pay-as-you-go meter (Azure-fakturering per forbruk).",
"disposition": "outdated"
},
{
"id": "ms-ai-security/cost-optimization/ai-builder-credits-transition.md#17",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/en-us/ai-builder/credit-management",
"evidence_quote": "By adding an AI Builder action to a flow inside your app, you turn your app into a premium app, so it requires a premium license to be used. ... Power Apps per app | 250 | Maximum = 1,000,000 AI Builder credits per tenant.",
"reason": "Første del holder (appen blir premium), men den lastbærende lisensdelen er for snever: kilden krever en premium-lisens generelt, og Power Apps per app er en premium-lisens som også gir AI Builder-kapasitet — det er ikke krav om nettopp Power Apps Premium.",
"file": "skills/ms-ai-security/references/cost-optimization/ai-builder-credits-transition.md",
"line": 309,
"claim": "En app som bruker AI Builder-funksjoner blir «premium», og brukeren som kjører appen må ha Power Apps Premium-lisens.",
"disposition": "outdated"
},
{
"id": "ms-ai-security/cost-optimization/ai-builder-credits-transition.md#21",
"judge_verdict": "not_grounded",
"rule": "R8",
"evidence_url": "https://learn.microsoft.com/en-us/ai-builder/endofaibcredits",
"evidence_quote": "Existing seeded customers will retain access to their AI Builder credits from seeded premium licenses through November 1, 2026, unless covered under a contract that started prior to November 1, 2025. In this case, seeded entitlements are honored through the contract term.",
"reason": "Datoen og at seeded credits fjernes også for EA-kunder stemmer, men den lastbærende delen «uten unntak» motsies: kontrakter inngått før 1. november 2025 honoreres ut kontraktsperioden.",
"file": "skills/ms-ai-security/references/cost-optimization/ai-builder-credits-transition.md",
"line": 385,
"claim": "Seeded AI Builder credits er inkludert i Enterprise Agreement (EA)-lisenser fram til 1. november 2026, og fjernes da også for EA-kunder uten unntak.",
"disposition": "outdated"
}
]
},
{
"file": "skills/ms-ai-security/references/cost-optimization/azure-ai-foundry-cost-governance.md",
"batch": "R7.1",