Supplier and Third-Party AI Assessments

This record provides a documented assessment of third-party AI components and services used by SecureXcel AI Services, ensuring that applicable suppliers...

SecureXcel/AIMS/B-034Procedure721 words 3 min readOwner: GRC & Compliance Team
Sign in to download the Word documentSign in to download the PDFFree. You return straight to this page after you sign in.
Published as written. This is SecureXcel's own working document, shown with its document control and notices as they appear in the file. Adapt it to your organization, approve it through your own document control, and treat it as a starting point: it is not legal advice and does not make an organization compliant or certified.

Document control

Document ID: B-034

Author(s): GRC & Compliance Team

Version: 1.0

Date: 09-Nov-2026

Disclaimer:

This controlled document is intended for authorized SecureXcel use. It may not be reproduced, distributed, quoted, or disclosed outside its approved audience without prior authorization from the document owner.

Document TitleSupplier and Third-Party AI Assessments
Document IdSecureXcel/AIMS/B-034Current Version1.0Date of Release09-Nov-2026
ClassificationInternalAuthorGRC & Compliance Team
Reviewed byGRC & Compliance Lead
Approved byManagement RepresentativeAI Governance Committee (AIGC)
Designation

Revision History

VersionDateAuthorGRC & Compliance TeamGRC & Compliance LeadManagement Representative
1.030/09/2026GRC LeadGRC HeadAIGCInitial version

1. Purpose

This record provides a documented assessment of third-party AI components and services used by SecureXcel AI Services, ensuring that applicable suppliers are evaluated against defined security, compliance, and AI-related requirements. It serves as evidence of supplier due diligence and supports ongoing monitoring of third-party AI suppliers in accordance with the organisation’s supplier management process.

2. Scope

Covers third-party AI models and libraries (from the Approved Software Tools / Open-Source AI Models Register), the URL-reputation API integration named in the SecureXcel AI Services Impact Assessment Report, and security-testing suppliers. General enterprise software (approved source-code repository, approved change and workflow management platform) is out of scope of this AI-supplier record unless it directly supplies or processes data for an AI system.

3. Suppliers Assessed

Supplier / componentCategoryLicence / contract typeNotes
Meta — Llama 3.1 (Llama-3.1-8B-Instruct)Open-source modelLlama 3.1 Community LicenceBase model, internal pilot use only per the Approved Software Tools register; on-premises air-gapped inference, no data leaves SecureXcel infrastructure
Approved open-source embedding / language modelOpen-source modelApache 2.0Fine-tuned, on-premises inference
Scikit-learn — k-means, Isolation ForestOpen-source library3-clause BSDTrained in-house, on-premises CPU
XGBoostOpen-source libraryApache 2.0Trained in-house, on-premises CPU
Qwen — Qwen3-VL-8B-InstructOpen-source modelApache 2.0Trained in-house, on-premises CPU
Approved external URL-reputation / threat-intelligence serviceThird-party service (URL reputation lookup)Commercial API agreementIntegrated into the CTA (Call-to-Action) Engine for malicious-link detection, per AIIA A.10.3
CERT-In empanelled VAPT auditor(s)Security testing serviceEngagement per audit cycleProvides external annual VAPT per the AIIA security architecture; internal quarterly VAPT run in-house
Third-party takedown partner(s)Partner integration (Third-Party Takedown API Module)Partner agreement(s)Supplier name(s) and agreement references to be confirmed with Vendor Management and added at next register update

4. Completed Vetting Checklist

Applying the B-008 Appendix C checklist to each supplier above:

Vetting requirementApproved third-party foundation modelApproved open-source modelApproved external URL-reputation / threat-intelligence serviceCERT-In VAPT partner
1. Documentation commitment (A.6.2.7): model card, licence terms and known limitations availablePass — public model card and licencePass — public model card and licencePass — API documentation availableN/A — service, not a model
2. Data governance alignment (A.7.4, A.7.5): data quality and provenance traceablePass — no SecureXcel data sent to supplier (on-prem inference)Pass — no SecureXcel data sent to supplier (on-prem inference)Review — URLs submitted for lookup; confirm no message content beyond the URL is sentN/A
3. Responsibility allocation (A.10.2): liability for defects allocatedPass — open-source licence terms applyPass — open-source licence terms applyReview — confirm liability terms in the commercial API agreementReview — confirm liability terms in the engagement letter
4. Alignment with prohibited usesPass — internal pilot use only, no prohibited use identifiedPass — internal pilot use only, no prohibited use identifiedPass — fraud/phishing detection is the intended useN/A
5. Ongoing monitoring access (A.6.2.6): logs/telemetry availablePass — on-premises, full internal telemetryPass — on-premises, full internal telemetryReview — confirm API usage logs are retained per B32Pass — VAPT reports retained as audit evidence

“Review” items above are not failures; they are the specific points Procurement / Vendor Management confirms and closes at the next register update, which is due within 60 days of this issue.

5. Maintenance

  • Reviewed whenever a new AI-relevant supplier, model or API is added (also update the AI System Inventory, B28, and Model and Change Register, B30).
  • Owner: Procurement / Vendor Management, with the AI System Owner confirming technical accuracy.

6. Compliance with ISO 42001:2023

RequirementWhere metClause
Suppliers of AI components assessedWhole documentA.10.3
Responsibilities allocated between supplier and organisationSection 4, item 3A.10.2

More in AI lifecycle, suppliers and operations

Sign in to download the Word documentSign in to download the PDFFree. You return straight to this page after you sign in.