Often, yes. The EU AI Act does not care where a company is based. It applies when an AI system is put on the EU market, or when what the system produces is used in the EU. A Canadian or US company with European customers, users or job applicants should check where it stands.

Who the Act covers

The Act sets duties for two main roles. A provider builds an AI system, or has one built, and offers it under its own name. A deployer uses an AI system in its own professional work. One company can be both: a software firm that builds an AI feature into its product is the provider of that feature, and a deployer of the AI tools its staff use.

Article 2 brings in companies outside the EU in two ways. Providers are covered when they place AI systems or general-purpose AI models on the EU market, wherever they are established. And providers and deployers located outside the EU are covered when the output of their AI system is used in the EU.

In practice that catches, for example:

  • a Canadian SaaS company that sells software with AI features to customers in Europe;
  • a US online shop whose support chatbot answers visitors from the EU;
  • a recruitment tool that scores applicants for jobs based in an EU country;
  • an app on the European app stores that generates images or text.

It does not reach purely personal, non-professional use, AI built only for scientific research, or a company whose AI is used only outside the EU and whose output never ends up there.

Four levels of risk

Most of what the Act asks depends on what the AI system is used for, not on the technology behind it.

  • Prohibited practices. A short list of uses is banned outright, including social scoring, manipulation that causes harm, exploiting people’s vulnerabilities, untargeted scraping of facial images, and emotion recognition at work or in schools.
  • High risk. AI used in sensitive areas, listed in Annex III: hiring and managing workers, education, credit scoring and access to essential services, critical infrastructure, biometrics, law enforcement, migration and the courts. AI that is a safety component of products such as machinery or medical devices (Annex I) is high risk too. These systems carry the heaviest duties: risk management, data governance, technical documentation, logging, human oversight, accuracy and registration.
  • Transparency. Chatbots must tell people they are talking to an AI, generated content must be marked as such in a machine-readable way, and deepfakes must be disclosed (Article 50).
  • Everything else. Spam filters, recommendations and most business tools carry no new duties beyond AI literacy.

The dates, after the 2026 Digital Omnibus

The Act entered into force in August 2024 and applies in stages. In 2026 the EU adopted an amendment, the Digital Omnibus on AI, which entered into force on 27 July 2026 and moved the high-risk deadlines back. The transparency duties kept their original date.

DateWhat applies
2 February 2025Prohibited practices, and the duty to support AI literacy among staff
2 August 2025Obligations for providers of general-purpose AI models
2 August 2026Transparency duties under Article 50, and most of the rest of the Act
2 December 2026Machine-readable marking for generative AI systems that were already on the market before 2 August 2026
2 December 2027High-risk duties for the Annex III areas (hiring, education, credit and so on)
2 August 2028High-risk duties for AI in regulated products (Annex I)

So for a company with a chatbot, an AI writing tool or an image generator facing EU users, the deadline is not 2027. The transparency duties have applied since August 2026.

What the Omnibus changed for smaller companies

Besides the new dates, the amendment eased a few things for small and medium-sized enterprises and for small mid-caps:

  • they may provide the technical documentation for a high-risk system in a simplified form, on a template the Commission provides;
  • their quality management system can be proportionate to their size;
  • the AI literacy duty now asks companies to take measures to support their staff’s AI literacy, rather than to guarantee a particular level of it.

Fines

Using a prohibited practice can cost up to €35 million or 7% of worldwide annual turnover, whichever is higher. Breaking most other duties, the transparency rules included, can cost up to €15 million or 3%. Giving authorities incorrect or misleading information can cost up to €7.5 million or 1%. For SMEs the cap is whichever of the two figures is lower, and for small mid-caps the same applies to the 3% and 1% tiers.

Where Canada stands

Canada has no AI-specific federal law in force. The Artificial Intelligence and Data Act, part of Bill C-27, died when Parliament was prorogued in January 2025. AI is still covered by the laws that already apply to the data it uses: PIPEDA federally, and the private-sector privacy laws of Alberta, British Columbia and Quebec. Quebec goes furthest: under Law 25, a business that makes a decision about a person based only on automated processing of their personal information must tell them so.

For a Canadian company selling into Europe, the EU AI Act is therefore often the strictest AI rulebook it has to meet, and meeting it covers most of what Canadian customers and regulators are starting to ask for.

What to do first

  1. List every AI system you build, sell or use, including the AI tools staff use on their own.
  2. Mark which of them have users, customers or output in the EU.
  3. Sort each one into the four risk levels. Anything that helps decide about people deserves a second look.
  4. Fix transparency now: tell people when they are talking to a bot, and label generated images, audio and video.
  5. Ask your AI vendors what they do with your data and how they meet the Act.
  6. Write down who owns each system, and keep the records an auditor or customer would ask for.

Our AI Governance & Compliance service does this work with you, from the inventory to the documentation and the controls in the software. For a first pass on your own, see our AI compliance checklist.

This article is general information, not legal advice. Dates and figures were checked in November 2026 against the sources below; for decisions about your own systems, work with a lawyer who knows the Act.