Montro
EU AI Act timeline from prohibited practices in 2025 to high-risk obligations in 2027 and 2028

Your AI inventory under the EU AI Act

The EU AI Act sets obligations by risk level for organisations that build or use AI. Prohibited practices have been banned since February 2025, and obligations for high-risk systems in Annex III apply from 2 December 2027. As a deployer, you need to know which AI systems you use and where each one sits.

Montro keeps those systems in the same register as your suppliers, including AI switched on inside tools you already approved. Each change arrives with its source, and your team decides.

Framework

Where your AI systems sit under the Act

The Act sorts AI by risk, and what you owe also depends on whether you build or use it. Some practices are banned, high-risk systems carry the heaviest duties, and AI that interacts with people carries transparency duties. Most everyday AI has none. General-purpose AI models have separate rules.

Prohibited

Prohibited practices

Certain uses of AI are banned under Article 5, whoever provides the system.

High-risk

High-risk systems

As a deployer, you use them as instructed, assign human oversight, monitor how they operate and keep the logs. Some deployers must also assess the impact on fundamental rights.

Limited risk

Transparency obligations

People must be told when they are dealing with an AI system, under Article 50.

How Montro helps

An AI inventory your team decides on

Montro records the AI systems you use as a deployer in the same register as your suppliers. Each one is raised with its source, mapped to the Act, and classified by the person who owns it.

An AI system in the register with its proposed risk level, owner and decision

AI switched on inside approved tools

A supplier turns on an AI feature in a product you approved last year. Montro raises it as a change to the contract, not a guess, so the approval can be reviewed.

A proposed risk level, for you to confirm

Each system is matched to a likely category, from prohibited to minimal risk. Your owner confirms it and records why.

Regulatory Mapping

An AI service that processes personal data or supports a critical function also appears in your GDPR, DORA or NIS2 records, from the same entry.

Book a conversation

What you get

What makes the inventory hold up

Three things your team can point to when someone asks.

Ownership

A named owner for every AI system

The person responsible for human oversight is recorded against each system, so it’s clear who answers for it.

Documentation

Evidence kept with each entry

Contract terms, supplier confirmations and your team’s assessments sit against the system they relate to.

Decisions

A record of every call your team made

Who decided, on what evidence, and when. AI services nobody uses any more are re-tiered, not removed.

Your role under the Act

Provider or deployer: it decides what you owe

Most firms

You are usually the deployer

If you use an AI system under your own authority, the Act treats you as a deployer. The company that built and supplies it is the provider, and carries the heavier duties.

Watch for

When a deployer becomes a provider

Putting your name on a high-risk system, substantially modifying it, or changing its purpose can shift provider duties to you.

The record

Your role, written down per system

Montro records whether you are the deployer or provider for each AI system, and who confirmed it.

FAQs

Questions about the EU AI Act, and how Montro keeps your AI inventory.

The Act entered into force on 1 August 2024 and applies in phases: prohibited practices from 2 February 2025, general-purpose AI model rules from 2 August 2025, most remaining provisions including transparency from 2 August 2026, high-risk systems in Annex III from 2 December 2027, and high-risk systems in products covered by Annex I from 2 August 2028.

Annex III lists the areas: biometrics, critical infrastructure, education, employment, access to essential services such as credit, law enforcement, migration, and justice. AI used as a safety component in products covered by EU product law can also be high-risk. Montro proposes a category for each system, and your team confirms it.

No. What you owe depends on your role and the system’s risk level. Providers of high-risk systems produce the technical documentation; deployers assign oversight and keep the logs; minimal-risk systems carry no specific duties. Knowing which systems you use is how you tell which duties apply.

GDPR still applies whenever an AI system processes personal data, so the two run side by side. For some deployers, the fundamental rights impact assessment under the AI Act complements the data protection impact assessment already required by GDPR.

Under Article 99, up to €35 million or 7% of worldwide turnover for prohibited practices, up to €15 million or 3% for most other obligations, and up to €7.5 million or 1% for supplying incorrect information. For SMEs, the lower of the two figures applies.

Keep reading

Where the EU AI Act meets your other obligations

AI tools rarely sit under one rule. These pages pick up where this one stops.

DORA

Financial entities must list AI suppliers that provide ICT services in their DORA register of information.

Read more

NIS2

If you’re in scope of NIS2, AI services in your supply chain fall under its supply-chain security duties.

Read more

GDPR & RoPA

AI tools that process personal data need an entry in your record of processing too.

Read more

Third-party records are becoming decision logs. The firms that get this right will be the ones that can show who decided, on what evidence, and when.

  • Pen-test summary on request
  • Frankfurt hosting
  • Sample register
  • Expert Series guests
  • Cyber Ireland member

See it on your own suppliers

A read-only connection, about an hour to set up, and a named person who can say yes to a decision.

Priced to scope — never per seat, never per supplier. Upgrade only when the evidence earns it.

Montro AI discovery audit dashboard