T03 Vendor Due-Diligence GCC Addendum
Verification required. This is a working template, not a regulator-approved form. Where it names a regional instrument, the claim comes from secondary sources and is a pointer to check, not a confirmed legal requirement. This is not legal advice.
How to use: This is a blank form. Copy it and fill in the empty cells and blanks for your system. To copy it, follow the "Suggest an edit to this page" link in the footer to reach its Markdown source.
Add to the standard third-party security questionnaire. Wiki control: AI-CTRL-014. Regional drivers reported: CBUAE vendor diligence expectation, NCA third-party domain, NCSA AI guidance [Reported].
Vendor: __ Product: _ Date: Reviewer: ___
Answer Yes / No / Partial. Attach evidence. "Partial" needs a dated plan.
A. Identity and certification
| # | Question | Answer | Evidence |
|---|---|---|---|
| A1 | Legal entity, country of incorporation, parent | ||
| A2 | Information security certification held (name, scope, expiry) | ||
| A3 | AI management certification held, if claimed (name, scope, expiry) | ||
| A4 | Any local or national seal or registration claimed (name, issuer, date) |
B. Data location and handling
| # | Question | Answer | Evidence |
|---|---|---|---|
| B1 | Country of inference, storage, logs, backups, support access | ||
| B2 | Any processing outside the customer's jurisdiction? List | ||
| B3 | Sub-processors and fourth parties (name, country, role) | ||
| B4 | Is customer data used to train or improve models? | ||
| B5 | Retention of prompts and outputs; deletion on request and at exit | ||
| B6 | Encryption at rest and in transit; key control | ||
| B7 | Can the customer pin processing to an approved region? |
C. Model governance
| # | Question | Answer | Evidence |
|---|---|---|---|
| C1 | Model card or equivalent (intended use, limits) | ||
| C2 | Training data provenance and licence position | ||
| C3 | Bias testing: method, last date, results sharing | ||
| C4 | Arabic support: tested dialects and quality results | ||
| C5 | Material model changes: notice period and testing | ||
| C6 | Human oversight features and kill switch / tenant disable | ||
| C7 | Explainability features for decisions about people | ||
| C8 | Content safety controls and evidence |
D. Security of the AI system
| # | Question | Answer | Evidence |
|---|---|---|---|
| D1 | Prompt-injection and data-exfiltration testing | ||
| D2 | Agent and tool permission model | ||
| D3 | Penetration test date and scope | ||
| D4 | Vulnerability management SLAs | ||
| D5 | Logging and customer access to logs |
E. Incident and regulator matters
| # | Question | Answer | Evidence |
|---|---|---|---|
| E1 | Incident notice to customer within how many hours? | ||
| E2 | Support for customer's regulator notifications | ||
| E3 | Past regulatory action or material breach (last 3 years) | ||
| E4 | Regulator and audit access rights granted |
F. Exit and continuity
| # | Question | Answer | Evidence |
|---|---|---|---|
| F1 | Data return format and deletion certificate | ||
| F2 | Service continuity and fallback | ||
| F3 | Subcontractor failure handling |
G. Decision
| Risk rating | Low / Medium / High |
|---|---|
| Conditions | |
| Contract clauses required | training-use ban, residency commitment, sub-processor notice, incident SLA, audit access, model-change notice |
| Approver / date | |
| Next review |