Does the Act matter if we only use someone else’s AI?

Yes. Using AI at work can create duties even when another company built the tool. Your role and what you use it for decide which rules apply.

Did all high-risk requirements start in August 2026?

No. Rules for uses such as hiring and credit decisions apply from 2 December 2027. High-risk AI inside regulated products has until 2 August 2028.

What counts as high-risk AI?

Certain AI uses listed in the Act, such as recruitment, creditworthiness assessment and access to education or essential services, can be high-risk. Some AI in regulated products is also covered. Classification depends on the specific use and applicable exceptions.

What does the EU AI Act ask about AI literacy for staff?

Companies that build or use AI must take steps to help the people who work with it understand it. The AI Omnibus of 27 July 2026 softened this duty, which earlier asked for a sufficient level of AI knowledge.

What does the EU AI Act ban?

It bans AI uses that clearly threaten people’s safety or rights. Examples are social scoring, harmful manipulation and emotion recognition at work or in education. Most bans have applied since 2 February 2025.

Does the EU AI Act apply to companies outside the EU?

Yes, in some cases. It covers companies that offer AI tools in the EU, paid or free, wherever they are based. It also covers companies whose AI results are used in the EU.

EU artificial intelligence law

EU AI Act: which rules apply, and when

The EU Artificial Intelligence Act (EU AI Act) sets stricter rules for AI uses that could cause more harm to people. It covers companies using AI tools, not only those building them.

Review your AI duties See our AI Governance service

Type: EU regulation on artificial intelligence
Approach: Four risk levels, from banned to minimal
In force since: 1 August 2024
Staff AI skills: Company duty since 2 February 2025

Who it applies to

Companies building, supplying or using AI

Businesses that build AI systems or put them on the market.
Employers using AI tools at work, including tools bought from another company.
Importers, distributors and makers of products with AI built in.
Companies offering large AI models that can be used for many different tasks.

Not every tool carries the same duties. Your role, the tool’s purpose and possible exclusions all count.

Dates to plan for

EU AI Act dates by duty

1 August 2024
The EU AI Act entered into force.
2 February 2025
Bans on certain AI practices began, together with the company duty to support staff understanding of AI.
2 August 2025
Duties began applying to AI models designed for many different tasks.
27 July 2026
The AI Omnibus amendment softened the staff literacy duty and moved some high-risk deadlines later.
2 August 2026
Most of the Act’s rules began to apply.
2 December 2026
A ban starts on AI that creates or alters realistic sexual images or videos of a real, identifiable person without consent. It also covers AI that creates child sexual abuse material.
2 December 2027
High-risk rules for listed uses, such as hiring and credit decisions, apply.
2 August 2028
High-risk rules start for AI built into regulated products.

What it asks of you

Check each AI use separately

List your AI tools
Record what each tool does, who owns it, what data it uses and whose decisions it affects.
Determine your role and duties
Check whether a use is prohibited, high-risk or subject to other duties, then record why.
Help staff understand AI
Companies that build or use AI must take steps to help their staff understand it.
Put people in charge and tell users
Where required, put people in charge of checking results and tell users when they interact with AI.
Document and test each system
Keep the documents, test results and monitoring logs your role requires for each AI system.

How we help

Put AI Act requirements into everyday practice

We assess your AI uses, identify applicable duties and guide implementation with your teams. We establish responsibilities, approval processes and human oversight, supported by training and ongoing reviews.

AI GovernancePut AI governance into practice: clear responsibilities, controlled use and support with applicable EU AI Act requirements. Security Awareness & Management TrainingPractical cybersecurity training that helps staff recognise threats and leaders manage security risks. Secure AI & AutomationWe build AI agents and automate workflows with appropriate safeguards, or assess and improve the security of AI already in use.

Common questions

Related regulations and standards

ISO 42001
An international standard for setting the roles, rules and checks around the AI a company builds or uses. Certification helps you prepare for the Act, but the Act’s legal duties still need their own check.

Free 30-minute call

Find the duties for your AI tools

Describe your AI uses in a free 30-minute call with a senior specialist and decide which tools to check first.

Review your AI duties hello@guardera.tech