# The EU AI Act asks one question per AI system: what is it, and what do you owe?

The EU AI Act regulates AI system by system. Your duties depend on what each system does, where it is used and whether you built it or deploy it. XGRC® AI Governance records that decision for every system, scopes the obligations that follow and keeps the evidence that you meet them.

**Frameworks:** EU AI Act, ISO/IEC 42001, GDPR, POPIA

## Most organisations know the AI Act exists. Few can show where each system stands.

The Act does not ask whether your organisation uses AI responsibly in general. It asks, for each system, which category it falls into, what your role is and what that requires. Without a recorded answer, every audit, customer questionnaire and regulator query starts from zero.

- Nobody can say which AI systems are high risk, limited risk or out of scope
- Classification decisions made in meetings and never written down with the legal reasoning
- Teams unsure whether they are the provider or the deployer of a tool they bought and then changed
- Transparency, logging and oversight duties known in principle but not tracked per system

## From AI inventory to scoped, evidenced obligations.

- List every AI system in use, including tools bought in, built in house or switched on inside existing software.
- Check each system against the prohibited practices first, and stop there if one applies.
- Decide the risk tier: high risk, transparency obligations only, or minimal risk, and record the reasoning against the article.
- Confirm your role for each system: provider or deployer, and whether a change has made you the provider.
- Scope only the obligations that fit that tier and role, then record met, partly met or not met with evidence.
- Turn every gap into an owned, dated action, and re-classify when the system changes.

## How it works

- A guided 18-question EU AI Act decision tree that records every answer against its article
- Approved classifications locked and versioned, so a re-classification never overwrites the last one
- Requirements that scope themselves from the approved tier and role, so a limited-risk deployer sees a short list
- Evidence attached per requirement, with gaps raised as owned actions
- Change rules that send a system back for re-classification when its capability or purpose changes

## On the platform

**The AI Act inside the platform you already run** — Classification, compliance, impact, risk, oversight and incidents all sit on the same AI system record in XGRC® AI Governance. The same controls and evidence can answer ISO/IEC 42001, GDPR and POPIA, so one piece of proof is not filed four times.

## The same obligations, with a decision you can defend.

| Manual approach | With XGRC® |
| --- | --- |
| AI use described informally | Every system classified against its article |
| Risk tier argued in email, never recorded | Approved decisions locked and versioned |
| Every obligation of the Act read as if it applies | Only the obligations for your tier and role |
| Re-classification only when someone remembers | Changes trigger re-classification by rule |

## One checklist from prohibited practices to incident reporting.

- Scope and prohibited practices
- Risk classification and role
- Deployer and provider obligations
- Transparency, AI literacy and incidents

## Frequently asked questions

### Does the EU AI Act apply to organisations outside the EU?

It can. It applies to providers who place AI systems on the EU market wherever they are based, and to providers and deployers outside the EU when the output of their AI system is used in the EU. UK and South African organisations with EU customers, staff or users should check each system.

### What is the difference between a provider and a deployer?

The provider develops an AI system, or has it developed, and places it on the market under its own name. The deployer uses it under its own authority. A deployer becomes the provider if it puts its own name on a high-risk system, makes a substantial modification or changes its intended purpose so that it becomes high risk.

### Do we need to classify AI tools we bought rather than built?

Yes. Buying a tool changes your role, not whether the Act applies. Deployers of high-risk systems have their own duties, such as human oversight by competent people, monitoring, keeping logs and informing affected workers.

### How does XGRC® record a classification?

A guided 18-question decision tree records each answer against its article and produces a tier and role. Once approved, the result is locked. Re-classifying creates a new version, so you can always show what was decided, when and why.

## Related solution

- [AI Governance](https://xgrcsoftware.com/ai-governance)

---

Source: https://xgrcsoftware.com/use-cases/eu-ai-act-compliance
