Regulation · EU AI Act

Get ready for the EU AI Act.

We find every AI system you use, classify it under the EU AI Act, show which obligations apply to you, and build the evidence on your governed AI layer.

  • Inventory and risk classification of every AI system
  • Obligations mapped to your role: provider, deployer or both
  • Logs, oversight and testing evidence produced automatically
BannedHigh riskSTRICT RULESLimited riskTRANSPARENCYMinimal riskNO NEW RULES
CV screening assistant

The reality

Most companies don't know which AI rules apply to them.

The EU AI Act applies to almost every company that builds or uses AI in the EU. The obligations depend on what each system does and your role in it, and most companies have never mapped either.

No overview

  • Nobody has a full list of AI in use
  • Shadow AI tools adopted by teams
  • AI built into SaaS products you already pay for
  • Unclear who owns each system

You can't classify what you haven't found.

Unclear obligations

  • Is this system high-risk?Unclear
  • Are we provider or deployer?Unclear
  • Do transparency rules apply?Often
  • Which deadlines matter to us?Urgent

The same tool can mean very different duties.

No evidence

  • Where are the logs
  • Who can override the AI
  • How was it tested
  • What did we tell users

Compliance needs proof, not intentions.

Risk-based rules

Your obligations depend on the risk, and on your role.

The AI Act sorts AI systems into four risk levels. What you must do depends on the level and on whether you build the system or use it.

Banned

Unacceptable risk

Practices that threaten people's safety or rights are prohibited.

ExamplesEmotion recognition at work, social scoring, manipulative techniques
Strict requirements

High risk

Allowed, with risk management, documentation, logging, human oversight and conformity assessment.

ExamplesCV screening, credit scoring, access to education, critical infrastructure
Transparency

Limited risk

People must know they are dealing with AI, and AI-generated content must be marked.

ExamplesCustomer chatbots, AI-generated images, text and voice
No new obligations

Minimal risk

Most AI falls here. No specific rules beyond AI literacy for staff.

ExamplesSpam filters, recommendations, internal productivity tools

Provider

You develop an AI system or have one developed, and put it on the market or into service under your name. Most obligations sit here.

Deployer

You use an AI system in your business. You need AI literacy, human oversight and logs for high-risk use, and transparency towards people affected.

General-purpose AI

Providers of general-purpose AI models have their own rules on documentation, copyright and, for the largest models, systemic risk.

Key dates

The clock is already running.

Some obligations already apply. The high-risk rules follow in 2027 and 2028, after the Digital Omnibus moved the original dates.

Today
Entry into force

The AI Act became law.

In force
Bans and AI literacy

Prohibited practices banned; staff AI literacy required.

In force
General-purpose AI

Rules for GPAI model providers and the penalty regime apply.

In force
Transparency

Disclosure and labelling duties under Article 50.

In force
High-risk (Annex III)

Obligations for stand-alone high-risk systems, such as HR and credit.

Upcoming
High-risk (Annex I)

Obligations for AI in regulated products, such as machinery and medical devices.

Upcoming

Dates reflect the AI Act as amended by the Digital Omnibus on AI. Source: AI Act Explorer.

How we help

Requirements turned into controls that run.

Most AI Act duties need evidence that is produced continuously. Our governed AI layer creates it as part of normal operation.

RequirementHow we cover itWhere it runs
AI inventory and classificationArt. 6 · Annex IIIEvery AI system found through the gateway and registry, then classified by role and risk.01Gateway04Run
Risk managementArt. 9Ongoing risk assessment backed by structured testing and attack testing.System testingRed teaming
Data governanceArt. 10Personal data masked before it reaches a model; data sources documented.02Protect
Logs and technical documentationArt. 11–12Every request traced and logged; documentation built from real system data.03Observe
TransparencyArt. 13 · Art. 50Users told when they deal with AI, and AI-generated content labelled.01Gateway
Human oversightArt. 14Approval steps for high-impact decisions and agent actions, with the decision recorded.04Run
Accuracy, robustness and cybersecurityArt. 15Measured accuracy, attack testing and guardrails on every request.02ProtectRed teaming
AI literacyArt. 4Practical workshops so your teams use AI safely and know the rules.Workshops

How it works

Get ready in five steps.

From a list of tools to a documented, defensible position.

01

Register your AI systems

We find every AI system in use, including shadow AI and AI inside SaaS tools, and record owners and purpose.

AI inventory
02

Classify role and risk

For each system we determine your role and its risk level, and list the obligations and deadlines that apply.

Classification report
03

Run a gap analysis

We compare what you have today with what the AI Act requires, and prioritise the gaps by risk and deadline.

Prioritised gap list
04

Close the gaps

We put the controls in place on your governed AI layer: logging, oversight, transparency, testing and data protection.

Controls in place
05

Keep evidence current

Evidence is collected continuously and new AI systems go through the same process when they arrive.

Audit-ready evidence

The outcome

What you'll have.

A complete AI inventory

Every system classified by role and risk

Clear obligations

Per system, with the deadline that applies

Evidence on demand

Logs, tests and decisions ready for review

Ready before deadlines

A plan for 2026, 2027 and 2028

FAQ

Frequently asked questions.

Who does the EU AI Act apply to?

It applies to providers, deployers, importers and distributors of AI systems placed on the EU market or used in the EU. This includes companies outside the EU when the output of their AI system is used in the EU.

Does it apply if we only use tools like ChatGPT or Copilot?

Yes. As a deployer you must ensure your staff have sufficient AI literacy. If you use AI in a high-risk area, such as screening job applicants, you also have deployer obligations such as human oversight and keeping logs.

What are the risk categories?

Unacceptable risk (banned), high risk (strict requirements), limited risk (transparency duties) and minimal risk (no new obligations). General-purpose AI models have their own set of rules.

When do we need to comply?

Bans and AI literacy apply since 2 February 2025, general-purpose AI rules since 2 August 2025 and transparency duties since 2 August 2026. High-risk obligations apply from 2 December 2027 for stand-alone systems and 2 August 2028 for AI in regulated products.

What are the penalties?

Up to €35 million or 7% of worldwide annual turnover for prohibited practices, up to €15 million or 3% for most other breaches, and up to €7.5 million or 1% for supplying incorrect information. For SMEs, the lower of the two amounts applies.

Do you replace legal advice?

No. We cover the technical side: inventory, classification support, controls and evidence. We work alongside your legal counsel, who should confirm your final legal position.

This page is general information, not legal advice.

Contact us

Know where you stand under the EU AI Act.

A 30-minute call with an engineer. We'll look at the AI you use today and what it means for your obligations.

Cookie settings

Manage what we store on your device. Read the full Cookies Policy.

Strictly necessary

Required for the site and its forms to work. Always on.

Always on

Analytics

Google Analytics & Hotjar — traffic measurement and anonymised session insights.