Internal · Confidential
Vandermeer ConsultingVandermeer Consulting NV · Brussels
Policy document

AI
Use Policy

How we use AI at Vandermeer Consulting: what's approved, what's off-limits, and who signs off before anything reaches a client. Aligned to the EU AI Act.

Vandermeer Consulting NV·Brussels·EU-hosted & GDPR-aligned

Version
1.0
Effective
17 June 2026
Next review
17 December 2026
Owner
AI Officer
AI Use Policy · the rules
Vandermeer Consulting · v1.0 · part 2 of 3
Purpose & scope why this exists

1 Purpose, scope, validity

Keeps Vandermeer Consulting compliant with the EU AI Act and protects the trust our community places in us. Applies to all staff, contractors and interns, and to AI built into tools we already run.

2 Three things every Vandermeer Consulting employee must do

01 · TRAIN
Complete AI literacy
Finish the 30-minute AI literacy module within 30 days of joining; refresh yearly.
EU AI Act, Art. 4 · in force Feb 2025
02 · STAY ON LIST
Use only approved tools
Run company work through the Approved AI Tools list (clause 5). No personal accounts, no shadow AI.
See clauses 4 and 5
03 · REPORT
Flag incidents same day
If AI output is wrong, biased, harmful, leaked or risky, report to the AI Officer before close of business.
See clause 9
Banned, full stop red lines

3 Prohibited uses

Banned · EU AI Act, Article 5

No social scoring of staff, customers or users.
No emotion or mood detection in the workplace.
No manipulative or subliminal AI techniques.
No biometric scraping or mass surveillance.
No deepfakes of real people without consent.

Banned · Vandermeer Consulting house rules

No AI as sole decision-maker for hiring, performance, promotion or termination. A human always decides.
No client, employee or financial data in a tool that isn't approved.
No personal accounts (free ChatGPT, free Gemini, etc.) for company work.
No AI-generated client deliverable shipped without human review.
Full prohibited list → EU AI Act, Article 5 · artificialintelligenceact.eu/article/5
Data & approved tools what goes where

4 What you can use with what data

Data type
Where you can use AI
Public information
Any approved tool.
Internal, non-sensitive
Approved tools only. No consumer or free-tier tools.
Customer, employee, financial
Tier 1 enterprise tools only, with DPA on file.
Source code, IP, trade secrets
Tier 1 only, with manager or AI Officer approval per use.

5 Approved AI tools ( Vandermeer Consulting register)

airlock is the only approved proxy connector. Every other connector sits behind it: one URL, policy set once, every call audited.

Tool
Tier
Access via
airlock · proxy connector
Required
Sophie Janssens, AI Officer
Claude Teams / Pro
Tier 1
via airlock
Live register: intranet/ai-tools · request new connectors via the AI Officer.
Vandermeer Consulting NV · Vandermeer Consulting · ai-policy-v1.0 · part 2 of 3 Built withairlock
AI Use Policy · the duties
Vandermeer Consulting · v1.0 · part 3 of 3
Transparency tell people

6 Always tell people when AI is involved

A chatbot's first message says I am an AI assistant. AI-generated images, audio or video that could be mistaken for real carry an AI-generated label. (EU AI Act Art. 50 · due 2 Aug 2026.)

A candidate, partner or user affected by an AI-supported decision is told, and offered a human review on request.

7 You stay accountable for what you ship

AI errors are your errors.

AI is a draft. You verify before sending, publishing, deploying or signing off. “The AI did it” is not a defence at Vandermeer Consulting.

Confidentiality & incidents handle with care

8 Confidentiality and IP

Do not paste anything into AI that you would not post publicly, unless the tool is Tier 1. Confidential business documents are Tier 1 by default.

AI outputs are Vandermeer Consulting property where the law allows. Tell your manager when a deliverable is AI-assisted.

9 Incidents, same business day

Output that's wrong, biased, harmful or leaked, or that could embarrass Vandermeer Consulting or harm someone, goes to the AI Officer before close of business, with what you did, what it produced and what data was involved.

Sanctions & contacts when in doubt, ask

10 Sanctions

Breach of this policy is treated under the Vandermeer Consulting disciplinary code, up to and including termination, and referral to authorities where required by law.

11 Where to ask

AI Officer
Sophie Janssens
ai-officer@vandermeer-consulting.example
Slack: @sophie · channel #ai-help
Data protection (DPO)
Geert Vandenbroeck
dpo@vandermeer-consulting.example
Until appointed, route to the AI Officer.

When in doubt, ask first. It is always cheaper than a leaked deck or a regulator's letter.

12 Acknowledgement & review

By continuing to use AI tools for Vandermeer Consulting work, you acknowledge this policy. Owner: AI Officer · Version 1.0 · Effective 17 June 2026 · Next review 17 December 2026, or sooner after any incident or significant regulatory change.
Anouk Vandermeer
Chief executive officer
Sophie Janssens
AI Officer
Vandermeer Consulting NV · Vandermeer Consulting · ai-policy-v1.0 · part 3 of 3 Built withairlock