04
AI Governance & Safety Compliance
AI Governance, engineered.
A Brainbox framework to run AI safely, legally and usefully, with the EU AI Act as the first step.
We combine the EU AI Act, QMS integration and the engineering to implement it.
What is AI governance?
The system of accountability, controls and evidence that keeps every AI system trustworthy, lawful and worth its cost, from first idea to retirement.
Accountability
Clear owners and decision rights for every AI system.
Risk control
Harms to people, safety and business identified and reduced.
Lifecycle
Gates from intake to build, launch, monitoring and retirement.
Evidence
Documentation and logs that prove it to clients and regulators.
Done well, governance doesn't slow AI down. It is what lets a company say yes, faster.
EU AI Act
The AI Act is one standard inside governance, and our way in.
Know
Classify each system by risk tier and your role, provider or deployer.
Own
Accountable owner, AI literacy for staff (Article 4).
Agree
No banned uses (Article 5); transparency for chatbots and generated content (Article 50).
Check
High-risk duties: risk management, logging, human oversight, documentation.
What applies, and when
- In force
Prohibited practices banned; AI literacy begins
- In force
General-purpose AI model obligations; supervisors designated
- In force
Transparency (Article 50); enforcement and penalties
- In 1 month
Machine-readable marking for systems already on the market; added prohibitions
- In 12 months
High-risk obligations for Annex III systems
- In 20 months
High-risk obligations for AI in regulated products (Annex I)
Dates reflect the Digital Omnibus, Regulation (EU) 2026/1744. High-risk dates are deferred, not cancelled: evidence for a high-risk system takes 12–18 months to build. Fines go up to €35 million or 7% of worldwide turnover.
Which role you hold
Provider
You develop an AI system, or have one developed, and place it on the market or put it into service under your own name.
Deployer
You use an AI system under your own authority in your organisation. Most businesses are deployers.
Importer or distributor
You place a non-EU provider's system on the EU market, or make systems available in between.
Outside the EU
The Act can still apply if the output of your AI system is used in the EU.
How to start
Start with a free 30-minute readiness call
- Bring one AI system or one question
- We map where AI sits and what likely applies
Next step after the call: the AI Exposure Check, an inventory of your AI systems with a risk and role classification and an evidence plan.
Why Brainbox: compliance, engineering and change in one team
Engineers, not auditors
We assess the real system: code, data, models.
Evidence by design
Logging, tests and decision records built into the pipeline.
Risk to return
We can ship a working, compliant AI use case.
People side built in
Two decades of change leadership, so governance gets adopted.
Fixed price, SME-sized
Clear scope and cost up front.
Local, with scale
Brainport-based, delivery in the Netherlands and India, specialists on demand.
Where this applies
EU AI Act readiness
Classification, owners, policies, evidence: the full loop built for the systems you run or build.
AI in safety-critical products
Governance that connects to functional safety practice (ISO 26262, IEC 61508) instead of duplicating it.
Supplier & customer AI questionnaires
Answer customer AI questionnaires with evidence, and ask the right questions of your own suppliers.
ISO/IEC 42001 preparation
Build the AI management system that certification audits expect, on top of the QMS you already run.
Free PDF guide
What the EU AI Act expects of your business
A plain-language guide for business leaders. Free PDF.
- Your role: provider, deployer, importer, or outside the EU
- The dates: what applies, and when
- What a compliant business can show
Frequently asked questions
Do we need to act if we only use tools like ChatGPT or Copilot?
Yes. As a deployer you have duties even when you only use AI: AI literacy for staff (Article 4), no banned uses (Article 5) and transparency where people interact with AI (Article 50).
What is AI literacy under Article 4?
Providers and deployers must ensure their people have a sufficient level of AI literacy: understanding what the system does, its limits and its risks. It applies since 2 February 2025.
What does Article 50 require?
Transparency: people must be told when they interact with an AI system such as a chatbot, and AI-generated or manipulated content must be labelled and machine-readable.
Is this legal advice?
No. We are engineers and governance specialists; legal interpretation is provided with a partner law firm [coming soon].
Do you work with companies outside the EU?
Yes. The Act applies based on where the output of an AI system is used, not only where a company is based.
